PREPARATION OF BaxSr1-xTiO3 THIN FILMS BY CHEMICAL SOLUTION

DEPOSITION AND THEIR ELECTRICAL CHARACTERIZATION

A THESIS SUBMITTED TO THE GRADUATE SCHOOL OF NATURAL AND APPLIED SCIENCES

OF THE MIDDLE EAST TECHNICAL UNIVERSITY

BY

UMUT ADEM

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE

IN THE DEPARTMENT OF METALLURGICAL & MATERIALS

ENGINEERING

DECEMBER 2003

Approval of the Graduate School of Natural and Applied Sciences

____________________________ Prof. Dr. Canan ÖZGEN

Director I certify that this thesis satisfies all the requirements as a thesis for the degree of Master of Science.

____________________________ Prof. Dr. Bilgehan ÖGEL Head of Department This is to certify that we have read this thesis and that in our opinion it is fully adequate, in scope and quality, as a thesis for the degree of Master of Science. ____________________________

Prof. Dr. Macit ÖZENBA� Supervisor Examining Committee Members Prof. Dr. Muharrem T�MUÇ�N ____________________________ Prof. Dr. Macit ÖZENBA� ____________________________ Prof. Dr. Ali KALKANLI ____________________________ Prof. Dr. Ra�it TURAN ____________________________ Asst. Prof. Dr. Mustafa GÖKTEPE ____________________________

iii

ABSTRACT

PREPARATION OF BST THIN FILMS BY CHEMICAL SOLUTION

DEPOSITION AND THEIR ELECTRICAL CHARACTERIZATION

Adem, Umut M.S., Department of Metallurgical & Materials Engineering

Supervisor: Prof. Dr. Macit Özenba�

December 2003, 98 pages

In this study, barium strontium titanate (BST) thin films with different compositions

(Ba0.9Sr0.1TiO3, Ba0.8Sr0.2TiO3, Ba0.7Sr0.3TiO3, Ba0.5Sr0.5TiO3) were produced by

chemical solution deposition technique. BST solutions were prepared by dissolving

barium acetate, strontium acetate and titanium isopropoxide in acetic acid and adding

ethylene glycol as a chelating agent and stabilizer to this solution, at molar ratio of

acetic acid/ethylene glycol, 3:1. The solution was then coated on Si and

Pt//Ti/SiO2/Si substrates at 4000 rpm for 30 seconds. Crack-free films were obtained

up to 600 nm thickness after 3 coating – pyrolysis cycles by using 0.4M solutions.

Crystal structure of the films was determined by x-ray diffraction while

morphological properties of the surface and the film-substrate interface was

examined by scanning electron microscope (SEM).

Dielectric constant, dielectric loss and ferroelectric parameters of the films were

measured. Sintering temperature, film composition and the thickness of the films

iv

were changed in order to observe the effect of these parameters on the measured

electrical properties.

The dielectric constant of the films was decreased slightly in 1kHz-1 MHz range. It

was seen that dielectric constant and loss of the films was comparable to chemical

solution deposition derived films on literature. Maximum dielectric constant was

obtained for the Ba0.7Sr0.3TiO3 composition at a sintering temperature of 800˚C for

duration of 3 hours. Dielectric constant increased whereas dielectric loss decreased

with increasing film thickness.

BST films have composition dependent Curie temperature. For Ba content greater

than 70 %, the material is in ferroelectric state. However, fine grain size of the films

associated with chemical solution deposition and Sr doping causes the suppression of

ferroelectric behaviour in BST films. Therefore, only for Ba0.9Sr0.1TiO3 composition,

slim hysteresis loops with very low remanent polarization values were obtained.

Keywords: BST, Pt/Ti/SiO2/Si substrate, DRAM, Chemical Solution Deposition.

v

ÖZ

BST �NCE F�LMLER�N K�MYASAL ÇÖZELT�DEN B�R�KT�RME

YÖNTEM�YLE HAZIRLANMASI VE ELEKTR�KSEL

KARAKTER�ZASYONU

Adem, Umut M.S., Metalurji ve Malzeme Mühendisli�i Bölümü

Tez Yöneticisi: Prof. Dr. Macit Özenba�

Aralık 2003, 98 sayfa

Bu çalı�mada, de�i�ik bile�imlerde (Ba0.9Sr0.1TiO3, Ba0.8Sr0.2TiO3, Ba0.7Sr0.3TiO3,

Ba0.5Sr0.5TiO3) baryum stronsiyum titanat (BST) ince filmler kimyasal çözeltiden

biriktirme yöntemiyle üretildi. BST çözeltileri, baryum asetat, stronsiyum asetat ve

titanyum izopropoksiti asetik asit içinde çözdürerek ve kararlıla�tırıcı olarak 3:1

asetik asit/etilen glikol mol oranında etilen glikol eklenmesiyle hazırlandı. 0.4M

çözelti kullanılarak, 3 kaplama-piroliz döngüsü sonunda, 600 nm kalınlı�a kadar

çatlaksız filmler elde edildi.

Filmlerin kristal yapıları x-ı�ını kırınımı ile belirlenirken, yüzeyin ve film-altlık

arayüzeyinin morfolojik özellikleri taramalı elektron mikroskobu ile incelendi.

Filmlerin dielektrik sabiti, dielektrik kaybı ve ferroelektrik parametreleri ölçüldü.

Sinterleme sıcaklı�ı, filmlerin bile�imi ve film kalınlı�ı bu parametrelerin ölçülen

elektriksel özelliklere etkisini gözlemlemek amacıyla de�i�tirildi.

vi

Filmlerin dielektrik sabiti 1kHz-1MHz aralı�ında yava�ça azaldı. Filmlerin dielektrik

sabitleri ve kayıplarının, literatürdeki di�er kimyasal solüsyondan biriktirme ile

hazırlanan filmlerle kar�ıla�tırılabilinir oldu�u görüldü. En yüksek dielektrik sabiti

Ba0.7Sr0.3TiO3 kompozisyonu için 800˚C’de 3 saat sinterlenerek elde edildi.

Dielektrik sabiti, film kalınlı�ıyla artarken dielektrik kaybı azaldı.

BST filmler bile�ime ba�lı Curie sıcaklı�ına sahiptirler. Ba miktarı % 70’ten fazla

oldu�unda, malzeme ferroelektrik yapıdadır. Fakat, kimyasal solüsyondan

biriktirmeye ba�lı ince tane boyu ve Sr katkılaması ferroelektrik davranı�ı

bastırmaktadır. Bu yüzden, sadece Ba0.9Sr0.1TiO3 bile�imi için, dü�ük artık

polarizasyonlu dar polarizasyon döngüleri elde edildi.

Anahtar Sözcükler: BST, Pt/Ti/SiO2/Si altlık, DRAM, Kimyasal Çözeltiden

Biriktirme.

vii

To my parents

viii

ACKNOWLEDGEMENTS

I am grateful to my supervisor Prof. Dr. Macit Özenba� for his guidance, impatience

and understanding throughout my thesis.

I would like to thank to Dr. �brahim Çam for his continuous help during the dielectric

and ferroelectric measurements and for his comments. I also would like to thank to

technical staff of Metallurgical Enginering Department, especially, Mr. Necmi Avcı

for x-ray diffraction studies and Mr. Cengiz Tan for SEM studies.

I cannot pay my debt to my lab friends: Tuna, whose friendship and knowledge, I

admire and acknowledge and Altan who inspired me to add more fun and sports to

research and life.

Finally, I would like to thank to all my friends who were with me: Galip, Ertu�,

Aytaç, Gül and all.

ix

TABLE OF CONTENTS

ABSTRACT .......................................................................................................... iii

ÖZ............................................................................................................................v

DEDICATION……………………………………………………………………..vii

ACKNOWLEDGEMENTS................................................................................ viii

TABLE OF CONTENTS...................................................................................... ix

LIST OF TABLES................................................................................................ xi

LIST OF FIGURES ............................................................................................ xiii

CHAPTER

1. INTRODUCTION...............................................................................................1

2. LITERATURE SURVEY ...................................................................................4

2.1. Dielectric and Ferroelectric Properties of Materials ..........................................4

2.1.1. Dielectric Properties ................................................................................4

2.1.2. Polarization .............................................................................................8

2.1.3. Ferroelectric Phenomena .........................................................................8

2.2. Ferroelectric Materials ...................................................................................12

2.2.1. Ferroelectric Materials...........................................................................12

2.2.2. Hysteresis Loops ...................................................................................12

2.2.3 Barium Strontium Titanate System .........................................................14

2.2.4. Applications of BST Thin Films ............................................................20

2.2.4.1 BST Films in DRAMs .....................................................................21

2.3. BST films on Si and Pt/Ti/SiO2/Si substrates..................................................24

2.4. Chemical Solution Deposition Technique.......................................................26

2.4.1. Process Chemistry .................................................................................26

x

2.4.2. Spin Coating Process .............................................................................29

2.4.3. BST Solution Preparation ......................................................................31

2.4.3.1. Solvents and Precursors ..................................................................31

2.4.3.2. Drying, Firing, Annealing of BST films..........................................31

3. EXPERIMENTAL PROCEDURE...................................................................34

3.1. Solution Preparation.......................................................................................34

3.1.1. Starting Materials ..................................................................................34

3.1.2. Solution Preparation ..............................................................................35

3.2. Coating of Films on Substrates.......................................................................40

3.2.1. Preparation of substrates.......................................................................40

3.2.2. Spin Coating..........................................................................................41

3.2.3. Heat Treatment of the Films ..................................................................42

3.3. Characterization of Films ...............................................................................43

3.3.1. Structural and Morphological Analysis ..................................................43

3.3.2. Dielectric and Ferroelectric Measurements……………………………..44

4. RESULTS AND DISCUSSION ........................................................................48

4.1. Improvement of Film Quality ........................................................................49

4.2. Dilution of the Solution..................................................................................52

4.3. Crystalline Film Formation ............................................................................53

4.4. Morphology of the BST Films........................................................................60

4.5. Dielectric Properties of BST Films.................................................................68

4.6. Ferroelectric Properties of BST Films ............................................................83

5. CONCLUSIONS AND FURTHER SUGGESTIONS......................................87

REFERENCES .....................................................................................................90

APPENDIX................................................................................................................97

xi

LIST OF TABLES

TABLE

2.1. Advantages and limitations of processes for producing BST thin films

[10, 11]. ..................................................................................................................17

2.2. Major applications of BST films with relevant properties. ................................21

2.3. Relative permittivity values and charge storage densities for medium

dielectric constant materials [6]...............................................................................23

2.4. Different precursors and solvents for some sol-gel derived BST films. .............32

3.1. BST solutions with different compositions and their corresponding

polarization states. ..................................................................................................34

3.2. Specifications of Pt/Ti/SiO2/Si substrate (INOSTEK Korea). ...........................41

3.3. X-ray energies of elements of BST film and Pt/Ti/SiO2/Si substrate. ................44

4.1. Compositions of BaxSr1-xTiO3 films that were produced in this study

listed according to the substrates they were coated on. ............................................49

4.2. Thermal expansion coefficients of Pt, BaTiO3 and SrTiO3 [35, JCPDS]............51

4.3 Grain sizes of 1 �m thick BST films sintered at different temperatures. .............63

4.4. Dielectric properties of some sol-gel derived films with different

processing parameters. Properties of films that were deposited using other common

BST deposition tchniques are also provided to give an idea.....................................74

4.5. Change in dielectric properties with increasing film thickness, for

xii

different compositions. ...........................................................................................77

4.6. Dielectric Constant and Loss of BaxSr1-xTiO3 films………………...................80

4.7. Ferroelectric properties of Ba0.9Sr0.1TiO3 films on Pt/Ti/SiO2/Si

substrates. ...............................................................................................................86

xiii

LIST OF FIGURES

FIGURE

2.1. Charging current IC and loss current Il in a dielectric [4].....................................6

2.2. Schematic of (a) the polarization behaviour of a ferroelectric material

below its Curie point, (b) the change in permittivity at a ferroelectric-paraelectric

phase transition and (c) the polarization behaviour of a ferroelectric material above

its Curie point [6]....................................................................................................10

2.3. The displacement of Ti+4 ion, for PbTiO3, from the body center of the

cubic perovskite structure, resulting in distortion of cubic structure and formation of

tetragonal perovskite structure [7]. ..........................................................................11

2.4. Creation of net polarization by poling [7]. ........................................................11

2.5. Classification of crystals according to the symmetry and polarization

they possess [8].......................................................................................................13

2.6. Hysteresis loop for a PbTiO3 thin film [7]. .......................................................14

2.7. Effect of several isovalent substitutions on the ferroelectric-paraelectric

transition temperature of ferroelectric BaTiO3 [11]. ................................................15

2.8. Lattice parameter change and stability regions of ferroelectric and

paraelectric phases of BST with composition [10]...................................................18

xiv

2.9. Variation of dielectric constant with temperature for bulk and thin film

BST [17].................................................................................................................19

2.10. Four stages of spin-coating shown schematically [25]. ...................................29

3.1. Flow chart for solution preparation (first route) and coating of films. ...............36

3.2. Flow chart for solution preparation (second route) and coating.........................37

3.3. SEM micrographs showing the surface of electroded BST films, (a)×400

(b) ×16. ..................................................................................................................45

3.4. Schematic of the modified Sawyer-Tower circuit. ............................................47

4.1. SEM micrographs, (top) showing surface cracks of 3 layers

Ba0.5Sr0.5TiO3 film (500 nm) coated by using first solution: solution containing 2

propanol, (bottom) showing crack-free surface of 3 layers Ba0.5Sr0.5TiO3 film

(600nm) coated by using second solution: solution containing ethylene glycol, both

films were sintered at 800˚C for 3 hours..................................................................50

4.2. XRD spectra of Ba0.5Sr0.5TiO3 films on Si substrate sintered at 600˚C,

700˚C and 800˚C for 3 hours...................................................................................54

4.3. XRD spectra of (a) 5 layers, 1 �m thick Ba0.5Sr0.5TiO3 film on Si,

sintered at 800˚C (b) 5 layers, 1 �m thick Ba0.8Sr0.2TiO3 film on Si, sintered at

800˚C......................................................................................................................55

4.4. X-ray diffractograms for Ba0.5Sr0.5TiO3 thin films coated on

Pt/Ti/SiO2/Si substrates and sintered at 600˚C, 700˚C and 800˚C using first solution

route………………………………………………………………………………... 58

4.5. X-ray diffractograms for Ba0.5Sr0.5TiO3 thin films coated on different

Pt/Ti/SiO2/Si substrates and sintered at 700˚C, 750˚C and 800˚C using second

xv

solution route (using ethylene glycol)......................................................................59

4.6. X-ray diffractograms of Ba0.5Sr0.5TiO3, Ba0.7Sr0.3TiO3 and Ba0.9Sr0.1TiO3

films (from bottom to top) sintered at 800˚C for 3 hours. ........................................60

4.7. X-ray diffractograms of (110) peaks of Ba0.5Sr0.5TiO3, Ba0.7Sr0.3TiO3 and

Ba0.9Sr0.1TiO3 films (from bottom to top) sintered at 800˚C for 3 hours…………...61

4.8. Cross-section SEM micrograph of 2 layers Ba0.5Sr0.5TiO3 film on

Pt/Ti/SiO2/Si substrate, (top ×5000, middle ×10000, bottom ×20000). ..................64

4.9. Cross-section SEM micrograph of 3 layers Ba0.5Sr0.5TiO3 film on

Pt/Ti/SiO2/Si substrate, sintered at 800˚C for 3 hours. .............................................65

4.10. SEM micrograph of Ba0.5Sr0.5TiO3 film on Si substrate. ................................65

4.11. EDS spectra of Ba0.5Sr0.5TiO3 film on Si substrate, sintered at 800˚C for

3 hours....................................................................................................................66

4.12. EDS spectra of Si substrate. ...........................................................................66

4.13. EDS spectra of�6 layers Ba0.5Sr0.5TiO3 film on Pt/Ti/SiO2/Si substrate

...............................................................................................................................67

4.14. EDS spectra of Pt/Ti/SiO2/Si substrate. ..........................................................67

4.15. The change of dielectric constant with frequency for 3 layers

Ba0.7Sr0.3TiO3 film (t = 600 nm). .............................................................................68

4.16. The change of dielectric constant with frequency for 3 layers

Ba0.7Sr0.3TiO3 film (t = 600 nm). .............................................................................69

xvi

4.17. Change in dielectric constant (at 1 kHz) with different sintering

temperatures, for 3 layers Ba0.5Sr0.5TiO3 films. .......................................................70

4.18. Dielectric constant of BST films with different compositions as a

function of frequency for 600 nm films…………………………………………… 71 4.19. The decrease in the Curie temperature of BaTiO3 with Sr addition [11]

...............................................................................................................................72

4.20. Dielectric constant of BST films with different compositions as a

function of frequency for 400 nm films. ..................................................................72

4.21. Dielectric loss as a function of frequency for 600 nm (3 layers) BaxSr1

xTiO3 films..............................................................................................................75

4.22. Dielectric loss of Ba0.5Sr0.5TiO3 films as a function of frequency

sintered at different temperatures for 600 nm films. ................................................76

4.23. SEM micrograph of the surfaces of Pt/Ti/SiO2/Si substrate: (top) prior

to any thermal treatment, (middle) after annealing at 800˚C for 1 hour. White dots

are hillocks of Pt, (bottom) after annealing at 800˚C for 3 hours. White dots have

grown in size. .........................................................................................................81

4.24. C-V curve for a 3 layers Ba0.7Sr0.3TiO3 film. ..................................................84

4.25. C-V curve for a 3 layers Ba0.9Sr0.1TiO3 film. ..................................................84

4.26. Polarization vs electric field behaviour for 600 nm Ba0.9Sr0.1TiO3 film

………....................................................................................................................85

A.1. Polarization versus electric field behaviour for 600 nm Ba0.9Sr0.1TiO3

film at Vx = 3.5 V....................................................................................................97

A.2. Polarization versus electric field behaviour for 600 nm Ba0.9Sr0.1TiO3

film at Vx = 3 V.......................................................................................................98

xvii

A.3. Polarization versus electric field behaviour for 600 nm Ba0.9Sr0.1TiO3

film at Vx = 2.5 V....................................................................................................98

1

CHAPTER 1

INTRODUCTION

In the late 1980s, conventional processing methods for producing electroceramic

materials (powder compacting, tape casting or screen printing techniques followed

by sintering processes) were increasingly supplemented by thin film deposition

techniques such as sputtering, pulsed laser deposition (PLD), chemical solution

deposition (CSD), and chemical vapor deposition (CVD). During these deposition

processes the electroceramic materials are synthesized on a microscopic scale

without powder processing as an intermediate step, usually at temperatures much

below the typical sintering temperatures for bulk ceramics [1].

This trend of thin film production was motivated by two events. Firstly, the newly

discovered HTSC (high temperature superconductors) promised to show

considerably larger current densities if deposited as thin films. Secondly, the idea

originating in the late 1960s of ferroelectric-semiconductor memories was taken up

again by utilizing the experience gained with the highly advanced Si semiconductor

technology. These pioneering lines led to a broad research and development effort

worldwide in order to combine the large variety of functions offered by the

electroceramic thin films with the standard integrated circuit chips [1].

Ferroelectric thin films have been receiving renewed attention for electronic

applications since 1989. The initial interest was for ferroelectric non-volatile

memories (i.e. FeRAMs [ferroelectric random access memories]), and this was

2

followed by the development of ferroelectric thin films in the paraelectric phase, for

use as high permittivity dielectrics for DRAMs (dynamic random access memories).

Most recently, there has been a substantial interest in the use of high permittivity

oxides as a replacement of SiO2 as the gate dielectric in mainstream Si CMOS

(complimentary metal-oxide-semiconductor) devices [2].

For DRAMs, improvements in density, cost, speed, and immunity with scaling are

critical. Up to recent years, SiO2 has been used as the capacitor dielectric material for

information storage in dynamic random access memories (DRAMs). To compensate

for the charge storage density loss resulting from scaling (reducing the dimensions

thus the area of the capacitor), thickness of SiO2 layer has to be decreased

continuously. However, it was seen that SiO2 thickness couldn’t be reduced below a

critical value that is set by the start of quantum mechanical tunneling of electrons

through the dielectric. This tunneling creates a high leakage current that makes

charge storage impossible. So, to further increase the storage density, materials with

high dielectric constant are expected to be employed to increase the capacitance per

unit area of the storage cell while keeping the thickness thick enough to prevent

leakage current. Ta2O5 was the first such material introduced into production of

DRAMs, using metal-insulator-semiconductor (MIS) structures. Later, higher

dielectric constant (K) materials will be introduced, to increase the capacitance per

unit area. Finally, materials with K > 100 will need to be exploited. Barium

Strontium Titanate (BST) is such a material, but it has serious problems that make it

doubtful whether BST can effectively be utilized as the storage capacitor dielectric.

The most difficult problem is that it is very difficult to scale BST’s thickness into the

desired range without preventing high leakage [3]. The high dielectric constant, low

dielectric loss and lack of fatigue or aging problems are the major advantages of BST

thin films for DRAM applications.

BST thin films are synthesized by generic thin film deposition techniques: RF-

magnetron sputtering, pulsed laser deposition (PLD), metalorganic chemical vapor

deposition (MOCVD) and chemical solution deposition (CSD). CSD method offers

3

advantages like stoichiometry control, homogeneity, low sintering temperatures and

most importantly low capital cost, making it process of choice for scientific research

and development.

In this study, BST films were deposited by coating the BST solutions prepared by

chemical solution technique on Si and Pt/Ti/SiO2/Si substrates using spin-coating

technique, in order to measure dielectric properties and to check whether the

dielectric properties are promising for DRAM applications and possibly for uncooled

infrared detector and microwave monolithic integrated circuit applications. This

thesis is composed of five chapters. Chapter 1 presents an introduction to DRAMs

and the materials for this memory technology, underlining the need for a high

dielectric constant material like BST. Chapter 2 consists of information on theory of

dielectrics, material properties and application fields of BST, common deposition

methods for BST thin films and chemical solution deposition in particular. Chapter 3

describes the experimental procedure, mainly solution preparation, coating and heat

treatment of the BST films. Chapter 4 consists of the results on solution

characterization, crystallization studies, microstructural characterization and

dielectric measurements and discusses the results by correlating the properties to

microstructure and processing parameters. Conclusions and further suggestions are

presented in Chapter 5.

4

CHAPTER 2

LITERATURE SURVEY

2.1. Dielectric and Ferroelectric Properties of Materials

2.1.1. Dielectric Properties

Dielectric materials are insulators, as they do not conduct electricity. The dielectric

constant (relative dielectric constant or relative permittivity) is a measure of the

ability of a material to store charge relative to vacuum and is a characteristic property

of a material. Capacitance of a material is defined as charge stored in the material per

unit voltage applied:

C (Farad) = Q (Coulomb) / V (Volt) Eq. [2.1]

For a parallel plate capacitor having no dielectric between the plates, capacitance is

equal to

dA

C×

= 0ε Eq. [2.2]

where �0 is the permittivity of vacuum, A is the area of each plate, and d is the

distance between the plates. When a dielectric material is placed between the plates,

the capacitance changes to

5

dA

C×= ε Eq. [2.3]

Relative dielectric constant K (�r), is the ratio of permittivity of the medium to the

permittivity of free space, �/�0.

In addition to dielectric constant, three other properties are important in determining

the insulating characteristics of a material: Electrical resistivity (�), the dissipation

factor (tan�) and the dielectric strength.

� is simply a measure of the resistance that a unit cube of the material offers to

current flow in a given (dc) field. In an ac field, the electrical resistivity and

dielectric constant are related by the dissipation factor which measures the energy

loss per cycle (usually in the form of heat) from the material.

If a sinusoidal potential )exp(0 iwtVV = is applied to the dielectric, the charge must

vary with time, which constitutes a charging current, Ic:

Q = CV; so )]2/(exp[0 π+==== wtiwCViwCVdtdV

CdtdQ

IC Eq. [2.4]

The charging current in an ideal dielectric thus leads the applied voltage by /2

radians (90˚) [4].

In addition to the charging current associated with the storage of electric charge by

the dipoles, a loss current must also be considered for real dielectrics. The loss

current arises from two sources: (1) The long-range migration of charges, e.g., dc

Ohmic conduction and (2) the dissipation of energy associated with rotation or

oscillation of dipoles. The latter contribution to the dielectric losses is a consequence

of the charged particles having a specific mass and, therefore, an inertial resistance to

being moved. Electrical energy from the field is lost in the overcoming of this inertia

6

during polarization. The ac conduction from the inertial resistance and the dc

conduction are both in phase with the applied voltage

IC

I

�

Il

Figure 2.1. Charging current IC and loss current Il in a dielectric [4].

Consequently a loss current in the dielectric can be written as

VGGI acdcl )( += Eq. [2.5]

where G is the conductance in units of Siemens, mho, or ohm-1. The total current IT

for a real dielectric material is thus the sum of equations [2.4] and [2.5]:

VGGiwCIII acdclCT )( ++=+= Eq. [2.6]

As shown in Figure 2.1, the total current in a real dielectric is a complex quantity

which leads the voltage by an angle (90 - �) where � is called the loss angle [4].

An alternative way of expressing the concept of a real dielectric possessing both

charging and loss processes is to use a complex permittivity to describe the material:

εεε ′′−′= j* and kikk ′′−′== 0** / εε Eq. [2.7]

7

where ε ′ and ε ′′ are the real and imaginary parts of complex permittivity,

respectively whereas *k , k ′ and k ′′ are complex relative dielectric constant, real

part of complex relative dielectric constant and imaginary part of complex relative

dielectric constant, respectively [4]. Thus, the total current in the dielectric can now

be expressed in terms of the single material parameter k*, since

0* CkC ×= where C0 is the capacitance of vacuum, so Q = CV = k*C0V Eq. [2.8]

and )exp()( 000* iwtiwVCkikiwVCk

dtdV

CdtdQ

i ′′−′==== Eq. [2.9]

and thus VCkwVCkiwIT 00 ′′+′= Eq. [2.10]

The first term on the right-hand side of the equation [2.10] describes the charge

storage in the dielectric and k ′ is thus called the charging constant or often just the

dielectric constant and ε ′′ and k ′′ are referred to as dielectric loss factor and relative

loss factor, respectively. The loss tangent, loss angle, or dissipation factor, tan �, is

defined as

kk ′′′=′′′= //tan εεδ Eq. [2.11]

The dissipation factor represents the relative expenditure of energy to obtain a given

amount of charge storage. Commonly, tan �× 100, called as “percent dissipation” is

used [5]. The product, k ′′ or k ′ tan � is sometimes termed as the total loss factor, and

provides the primary criterion for evaluating the usefulness of a dielectric as an

insulator. The inverse of the loss tangent, Q = 1/ tan �, is used as a Figure of merit in

high frequency insulation applications.

8

2.1.2. Polarization

When a material is placed in an electric field, the atoms and the molecules of the

materials become polarized so that we have a distribution of the dipole moments in

the material. An electrical dipole moment, p, is simply a separation between a

negative and a positive charge Q of equal magnitude: p = Q× a, where Q is the

electrical charge and a is the vector from the negative to the positive charge. This

separation of positive and negative charges and the resulting induced dipole moment

are termed as polarization, P, which is equal to the total dipole moment induced in

the material by the electric field: P = N×pav, where N is the number of molecules per

unit volume and pav is the average dipole moment per molecule. The polarization, P,

induced in a dielectric medium when it is placed in an electric field depends on the

field itself. To express the dependence of P on the field, E, we define a quantity

called the electric susceptibility, �, so that

P = � �0 E. Eq. [2.12]

There are four primary mechanisms of polarization in ceramics and glasses. Each

mechanism involves a short-range motion of the charge and contributes to the total

polarization of the material. These are: electronic polarization, ionic (atomic)

polarization, dipolar (orientational) polarization and interfacial polarization. Of these

mechanisms, dipolar polarization is important for ferroelectrics. It is responsible for

dielectric constant values of 104 or more in such materials. Rather than rotating, the

dipoles align and reverse align by cooperative ion displacements in ferroelectrics [5].

2.1.3. Ferroelectric Phenomena

Crystals are divided into 32 classes of point groups according to the symmetry they

possess. Of these 32 groups 20 of 21 are non-centrosymmetric (lacking center of

symmetry) and develop potential difference when a stress is applied on them. These

are piezoelectric materials. 10 of 20 non-centrosymmetric groups have spontaneous

polar axis and this spontaneous polarization depends on temperature. These 10

9

groups are pyroelectric. Ferroelectrics are subset of pyroelectrics; their spontaneous

polarization is reversible by the application of electric field (Figure 2.2).

As an example, for PbTiO3, (Figure 2.3 shows the paraelectric and ferroelectric

phases and the displacement of Ti+4 ion from the equilibrium position in the

tetragonal phase below Curie temperature) the Ti+4 ion is displaced up or down from

the body center position in the unit cell, this distorted structure is stable up to a

temperature, 490˚C (120˚C for BaTiO3), which is so called Curie temperature of the

material, the temperature at which a paraelectric material goes into a phase transition

from a non-polar phase to polar ferroelectric phase. This displacement is the origin of

spontaneous polarization. The centers of negative and positive charges do not

coincide. Since the Ti ion is displaced from the equilibrium body center position with

respect to barium and oxygen ions there is a net polarization or dipole moment

(positive and negative charges separated by a distance) exists in the unit cell. The

alignment of dipole moments in the same direction along neighboring unit cells

creates the domains. The domains oriented in different directions are separated by

boundaries, called domain walls.

Above the Curie point, the temperature of phase transition, the dielectric permittivity

falls off with temperature according to the Curie-Weiss law (Figure 2.2)

00 TT

C−

+= εε ~ 0TT

C−

Eq. [2.13]

where C is the Curie constant, T0 (T0 ≤ Tc) is the Curie-Weiss temperature. In the

paraelectric regime, the spontaneous polarization is zero but the permittivity remains

high.

10

.

Figure 2.2. Schematic of (a) the polarization behaviour of a ferroelectric material

below its Curie point, (b) the change in permittivity at a ferroelectric-paraelectric

phase transition and (c) the polarization behaviour of a ferroelectric material above

its Curie point [6].

11

Figure 2.3. The displacement of Ti+4 ion, for PbTiO3, from the body center of the

cubic perovskite structure, resulting in distortion of cubic structure and formation of

tetragonal perovskite structure [7].

With no electric field, the net spontaneous polarization is 0 due to random orientation

of dipoles in the domains. But when you apply an electric field, or “pole” the

material, the domains are oriented in the same direction by domain wall movement

and the material will have a net spontaneous polarization as shown in Figure 2.4.

Figure 2.4. Creation of net polarization by poling [7].

12

2.2. Ferroelectric Materials

2.2.1. Ferroelectric Materials

Ferroelectric ceramics were born in the early 1940s with the discovery of the

phenomenon of ferroelectricity as the source of the unusually high dielectric constant

in the ceramic barium titanate capacitors. Since that time, they have been the heart

and soul of several multibillion dollar industries, ranging form high-dielectric

constant capacitors to later developments in piezoelectric transducers, positive

temperature coefficient devices, and electrooptic light valves. Materials based on two

compositional systems, barium titanate and lead zirconate titanate, have dominated

the field throughout their history. The more recent developments in the field of

ferroelectric ceramics, such as medical ultrasonic composites, high-displacement

piezoelectric actuators (Moonies, RAINBOWS), photostrictors, and thin and thick

films for piezoelectric and integrated-circuit applications have served to keep the

industry young amidst its growing maturity [8].

Figure 2.5 shows the classes of crystals according to the symmetry they possess.

Ferroelectric materials that are technologically important are ceramic perovskites that

constitute BaTiO3 (Barium Titanate), PZT (Lead Zirconate Titanate), PLZT (Lead

Lanthanum Zirconate Titanate), PT (Lead titanate), PMN (Lead Magnesium Niobate)

and solid solutions of elements that form the perovskite structure.

2.2.2. Hysteresis Loops

The ferroelectric materials give non-linear polarization response to the applied

electric field. Figure 2.6 shows the hysteresis behaviour of a ferroelectric film [7].

With the application of the electric field domains start to align with the applied E-

field, and polarization goes linear with the applied field. Then, further increase in the

electric field helps domain walls to move easily and polarization starts to increase

faster. When all the domains are oriented in the same direction the polarization

saturates, further application of the electric field will not make any difference. This

13

Figure 2.5. Classification of crystals according to the symmetry and polarization

they possess [8].

maximum polarization value is called saturation polarization, PS. When E-field is

removed, polarization drops to a certain value, Pr called remnant polarization;

polarization remained in the material when the applied field is 0. The application of

electric field in the opposite direction causes alignment of domains in the opposite

32 Symmetry Point

Groups

21 Noncentrosymmetric 11 Centrosymmetric

(Non-piezoelectric)

20 Piezoelectric

Polarized under stress

10 Pyroelectric

Spontaneously polarized

Subgroup Ferroelectric

Spontaneously polarized

Polarization reversible

Tungsten Bronze

Oxygen Octahedral

ABO3

Pyrochlore Cd2Nb2O7

Ceramic Perovskites

BaTiO3

PZT

PLZT

PT

PMN

(Na,K)NbO3

Layer Structure Bi4Ti3O12

14

direction and at a certain value of electric field called coercive field, (-Ec) the net

polarization in the material becomes zero. When the electric field is increased in the

reverse direction, polarization saturates again, this time in the reverse direction. Then

to complete the cycle, electric field is removed, material will have –Pr. Application

of electric field in the initial direction, brings the material to zero polarization state at

coercive field, Ec.

Figure 2.6. Hysteresis loop for a PbTiO3 thin film [7].

2.2.3 Barium Strontium Titanate System

BaTiO3 is a ferroelectric perovskite and has been well studied in bulk ceramic form

where the measured permittivities are well into thousands. BaTiO3 in bulk form is

still used as multilayer or discrete capacitors. It derives its high dielectric constant

from an ionic displacement and therefore differs from lower dielectric constant

materials such as SiO2, which experience an electronic displacement only with

changing applied voltage [9].

15

Various elements have been doped to ceramic BaTiO3 and the effect of alloying

elements on the ferroelectric-paraelectric transition temperature has been reported.

Figure 2.7 gives the effect of isovalent alloying elements on the Curie temperature

(ferroelectric-paraelectric transition temperature, Tc) of bulk BaTiO3. The utilization

of BaTiO3 - SrTiO3 solid solution allows the Curie temperature of BaTiO3 be shifted

from 120˚C to around room temperature for BaxSr1-xTiO3 films. For Sr addition into

BaTiO3, the linear drop of Tc is 3.4˚C per mol%. Therefore, 30 mol% Sr (x = 0.7)

would bring the Tc down to room temperature [10].

Figure 2.7. Effect of several isovalent substitutions on the ferroelectric-paraelectric

transition temperature of ferroelectric BaTiO3 [11].

BST as thin films have been under investigation for about 15 years starting from

90’s. The initial interest in BST thin films was due to its high dielectric constant, low

dielectric loss, high dielectric breakdown and composition dependent Curie

temperature enabling it to be in paraelectric state as well as ferroelectric state, all of

which makes it a candidate for replacing SiO2 as charge storage dielectric for

DRAMs (dynamic random access memories) [12, 13]. Recently, research on BST

thin films concentrates also on uncooled infrared detector applications [14, 15] and

16

on microwave monolithic integrated circuit applications [16], exploiting its

pyroelectric and electric field dependent (tunable) dielectric constant, respectively.

BST films are polycrystalline. Their properties heavily depend on composition,

stoichiometry, microstructure (grain size and grain size distribution), film thickness,

characteristics of the electrode material, and homogeneity of the film. The BST thin

film growth method significantly affects the composition, stoichiometry, crystallinity

and grain size of the film and, consequently, its dielectric properties. A variety of

techniques (Table 2.1) such as rf-sputtering, laser ablation, metalorganic chemical

vapor deposition (MOCVD), metalorganic deposition (MOD) and sol-gel processing

have been used to deposit BST thin films. Above methods are highly competitive,

each having advantages and disadvantages in terms of homogeneity, processing

temperature and processing cost. Because of the multicomponent nature of BST

films, precise microscopic control of stoichiometry is essential for obtaining uniform

single-phase films.

BaxSr1-xTiO3 films are not only paraelectric at the DRAM operating temperature

range (0-70˚C ambient and 0-100˚C on chips) but also achieve maximum

permittivity around the operating temperature. On the other hand, the volatilities of

the BST components are lower than the Pb-based ferroelectric materials, thereby

making it relatively easier to introduce into fabrication facilities [10]. Similar to

SrTiO3, it does not exhibit much frequency dependence of dielectric constant.

Barium-rich compositions of BST should be ferroelectric in theory, since BST is

below its Curie temperature at room temperature for Ba-rich compositons. For mol %

Sr � 0.25, the structure is ferroelectric tetragonal perovskite, for other compositions,

for % Sr 0.25, the structure changes to paraelectric cubic perovskite. Figure 2.8

shows the changes in the lattice parameter of BaxSr1-xTiO3 with composition. For

thin films, only a few researchers reported hysteresis loops as an evidence of

ferroelectricity. Cheng et al. explain the possible reasons for the disappearance of

ferroelectricity in sol-gel derived BST thin films as follows: [14]

17

Table 2.1. Advantages and limitations of processes for producing BST thin films

[10, 11].

METHOD ADVANTAGES LIMITATIONS

Chemical Solution

Deposition

• Inexpensive, low

capital investment

• Rapid sampling of

materials

• Homogeneity

• Low processing

temperatures

• Phase control

• Morphology

• Reproducibility

Pulsed Laser

Deposition

• Rapid sampling of

materials

• Quickly produce new

materials

• Morphology

• Point defect concentration

• Scalability (small areas

only)

• Uniformity

• High residual stress

Sputtering

• Cost

• Uniformity

• Scalability

• Standard IC

processing

• Low growth

temperatures

• Point defect concentration

• Residual stresses

• Stoichiometry control

• Slow deposition rate for

oxides

MOCVD

• Uniformity

• Morphology

• Small

• Scalability

• Immature technology

• Precursor stability

• Precursor availability

• Expensive

18

• It was observed that below a certain grain size called as critical grain size,

although the material is in ferroelectric state (it is below its ferroelectric-paraelectric

transition temperature), there was no anomaly observed for dielectric constant at the

transition temperature. The reasons for the disappearance of the anomaly is not

known precisely [7]. This critical size for the existence of ferroelectricity of BaTiO3

(around 120 nm) is much larger than that of PbTiO3 (around 7 nm). As an example,

Cheng et al. gives the research of Tahan et al. [13] whose films having grain sizes

ranging from 20 to 50 nm, and no ferroelectricity was present.

Figure 2.8. Lattice parameter change and stability regions of ferroelectric and

paraelectric phases of BST with composition [10].

• The sol-gel deposition of BaTiO3 films commonly results in polycrystalline,

granular films with grain diameters of lower than 70 nm due to random

nucleation in the pyrolyzed gel films. This is in contrast to Pb(Zr, Ti)O3 films,

which can easily be grown by the sol-gel method into a columnar or epitaxial

structure, because of heterogeneous nucleation from nanocrystalline pyrochlore

phase into the perovskite phase at the substrate.

19

• The tetragonality (c/a) of BST films decreases with the increasing strontium

content and a small value of c/a value may not be sufficient to create charge

separation and spontaneous polarization in the films, so the fabrication of BST

ferroelectric film is even more difficult than that of BaTiO3.

Figure 2.9 shows the variation of dielectric constant with temperature for bulk and

thin film Ba0.7Sr0.3TiO3. As can be seen from the graph, bulk BST undergoes a

maximum in dielectric permittivity around 310K (slightly above room temperature)

as addition of 30 % Sr to barium titanate shifts the Curie temperature of barium

titanate to room temperature from 120˚C. In contrast, thin film BST does not show a

peak or anomaly in dielectric permittivity due to the absence of ferroelectricity in

relation to the reasons above.

However, some researchers including Cheng, have been successful at fabricating

ferroelectric BST thin films. The main point in their succession was achieving to

fabricate BST films with columnar grains, having grain diameters greater than 100

nm, by coating very thin individual layers (< 10 nm [14]) using highly diluted

solutions.

Figure 2.9. Variation of dielectric constant with temperature for bulk and thin film

BST [17].

20

2.2.4. Applications of BST Thin Films

The primary electronic devices for which BST films investigated are (Table 2.2):

DRAM capacitor dielectrics, uncooled infrared detectors and monolithic microwave

integrated circuits.

Numerous reports are available in the literature involving the study of BST films for

DRAM applications as well as on the study of BST thin films for microwave

dielectric applications (such as phase shifters, varactors, delay lines etc). However,

more studies on BST for DRAM applications have been reported because for these

applications, synthesis of polycrystalline thin films is sufficient, whereas for

microwave dielectric applications oriented grain or epitaxially grown films are

required. The presence of grain boundaries increases the dielectric loss and may

consequently limit the performance [16].

In sol-gel processing the precursor solution is either spin or dip-coated on the

substrate at room temperature, in contrast, in almost all the other techniques (rf

sputtering, electron beam evaporation, chemical vapor deposition and laser ablation)

substrates are heated during film deposition. The hot substrates provide the adatom

mobility of the deposited species and also act as an effective nucleation site to

control the growth orientations of the films. When lattice matched substrates having

identical crystal structures and similar thermal expansion coefficients are used,

substrates promote epitaxial growth.

All the sol-gel deposited films are amorphous after deposition and needed to be post

annealed at relatively high temperatures for crystallization. Films become stressed

due to removal of organics during annealing. The other possible contributions of

stress are due to lattice parameter mismatch between film and substrate and also due

to change from amorphous to crystalline state. This intrinsic and extrinsic stress often

destroys the ordered growth influenced by single crystalline substrate. Therefore,

most of the sol-gel derived films are reported to be polycrystalline in nature [18].

21

Table 2.2. Major applications of BST films with relevant properties.

Application Composition Property Reference

DRAM (Dynamic

Random Access

Memory)

Ba0.5Sr0.5TiO3

(Paralectric)

High dielectric

constant, low

leakage current

[19]

Uncooled infrared

Detector

Ba0.85Sr0.15TiO3

(Ferroelectric)

High dielectric

constant and high

pyroelectric

coefficient

[15]

Phase Shifter Ba0.6Sr0.4TiO3

(Paraelectric)

Low dielectric

loss, high

tunability

[16]

2.2.4.1 BST Films in DRAMs

The main interest in producing high dielectric constant BST thin films is due to the

efforts of semiconductor industry about finding an alternative high dielectric constant

material for usage in DRAM’s (dynamic random access memories).

The DRAM is the primary working medium for information storage in the ubiquitous

microelectronic devices that comprise the entire litany of electronic systems. DRAM

works very simply by using a submicron-sized capacitor, representing one bit of

memory, to store a given amount of electrical charge: if the charge is present, it

represents a digital “1”; if not then the bit is a “0”. Each bit is addressed using a

CMOS-FET (Complimentary Metal Oxide Semiconductor Field Effect Transistor),

which acts as a valve for adding or removing the charge from the capacitor upon the

application of a voltage. It is accepted that DRAM is volatile, and the information

22

held by a DRAM is accessed at speeds approaching those at which the

microprocessor operates, while archival storage is handled by other means, such as

magnetic media or flash memory, that operate much more slowly. Information in a

DRAM must be refreshed by rewriting each bit on regular basis, as the charge in the

capacitor “leaks” away, causing volatility. This refresh time is typically 256 ms [12].

In order to increase the stored charge density, DRAM’s should scale down, meaning

that their size will be reduced. Scaling of DRAM’s results in shrinkage of SiO2/Si

capacitor area. Since, capacitance of the cell is directly proportional to the area, to

maintain the same charge storage density, the thickness of the dielectric SiO2 layer

should be decreased according to parallel plate capacitor equation: d

AKC

××= 0ε

,

where K is the relative dielectric constant, A is the area of the top electrode, �0 is the

permittivity of free space (8.85*10-12 Farad/m), and d is the thickness of the film.

But as SiO2 thickness is decreased below critical 10 nanometers, the leakage current

increases abruptly due to direct tunneling in the oxide layer. So scientists and

engineers realized that to maintain the same capacitance density (capacitance per

area), something other than reducing the thickness of oxide layer should be done.

These research efforts found two paths for solution to this scaling limit problem.

First path is the production of complex shaped capacitors that have larger area than

so called “planar capacitors”. But this solution increases the processing costs and it is

not easily applicable to Si technology. Second path is the replacement of SiO2 gate

dielectric with a higher dielectric constant material so that thicker oxides can be

placed on Si while maintaining the capacitance density and low leakage current

during scaling. For example SiO2 has a dielectric constant 3.9, replacement of SiO2

with Si3N4 (has a dielectric constant between 7 and 9) can maintain the same

capacitance density while having small enough leakage current due to prevention of

tunnelling current by increasing the physical thickness of the oxide [12].

23

In the recent years thin film perovskite materials with high dielectric constant such as

PZT, SrTiO3 and (Ba,Sr)TiO3 (BST) have been investigated as dielectric materials

for future DRAM’s. The best suited dielectric material would have a low leakage

current and a high dielectric constant and would also be in paraelectric phase to

avoid from fatigue from ferroelectric domain switching. SrTiO3 has a smaller

dielectric constant than BST and PZT is in ferroelectric phase at room temperature.

Thus, BST is very appealing for DRAM capacitors [10].

So, the research is mainly focused on high and medium dielectric materials. The

medium dielectric constant materials include Ta2O5, TiO2, ZrO2 and Y2O3 as main

candidates. The candidates other than BST have dielectric constants smaller than 40

(Table 2.3), but BST has a dielectric constant 200 (for a 20 nm film). This means

that BST can be exploited for more generations than Ta2O5 (tantalum pentoxide)

which is now in the production line.

Table 2.3. Relative permittivity values and charge storage densities for medium

dielectric constant materials [6].

Material Relative Permittivity C/A (fF/�m2)

Ta2O5 25 13.8 (20.4)

TiO2 30-40 9.3

ZrO2 14-28 9.9

Y2O3 17 4.7

Si3N4 (comparison) 7 7-8.6

But, BST has some disadvantages:

• The additional set of new materials required to isolate and integrate BST in

Si based chips.

24

• The thickness dependence of its dielectric constant.

• Like in most high-permittivity dielectrics, a significant time dependence

exists for the polarization response which can limit the fraction of stored charge that

can be read from a cell at the nanosecond timescales being approached by

commercial microelectronic devices. This places particularly stringent tolerances on

film quality, most conveniently measured in terms of dielectric loss tangent, which

must be less than 0.005 for the DRAM dielectric.

• BST failure mechanisms are poorly understood and entirely different from

those of SiO2.

Despite its disadvantages, BST is still the only candidate material nominated for

future use in DRAM cells starting from 2005 as declared in 2001 ITRS (International

Technology Roadmap for Semiconductors) report [3]. This fact is the motivation for

research and development in the field.

2.3. BST films on Si and Pt/Ti/SiO2/Si substrates

The interaction between a thin film and its substrate can be a very important factor in

determining the electrical properties of the film. When such a film is integrated into a

device, there is an interface between the film and the substrate, and between the film

and one or more electrodes deposited over the film. The device is usually subjected

to further processing, involving heating the system to several hundred degrees, so the

interactions between the film and the electrodes become more pronounced [20].

There are a number of criteria which affect the choice of the electrodes for thin

dielectric films [21]:

• The electrode must have sufficiently low resistance,

• It must be chemically compatible with the semiconductor, and not react with the

film,

• It must display adequate adhesion to both the film and underlying structure,

• It must be morphologically stable under the processing conditions,

• It may need to act as a diffusion barrier for oxygen and other species,

25

• It must provide suitable interfacial electronic properties,

• It may be necessary to utilize the electrode to control the film structure and

orientation (As an example, recently MgO substrate is used for epitaxial growth

of BST thin films instead of Pt/Ti/SiO2/Si substrates [16].

The traditional SiO2 or Si-O-N dielectrics are grown directly onto silicon (which itself

used as one capacitor electrode or plate), in a configuration known as a metal-insulator-

semiconductor (MIS) capacitor. But, because of reaction with silicon, this cannot be

adopted for the BST dielectric. Therefore a configuration was required in which the

BST was in contact with a different bottom electrode, in a metal-insulator-metal (MIM)

configuration. In addition, BST must be processed at relatively high temperatures and

oxygen partial pressures (to form the oxide structure), with the implication that the

electrodes must be stable against oxidation, and with the additional requirement for an

oxygen barrier under the electrode to ensure that the contact (termed the plug) down to

the underlying access transistor is not oxidized. Thus, the electrode material below BST

should be a noble metal or a conductive oxide [12].

Gold and silver react with the perovskites during processing, having solid solubility

in the perovskite lattice. Therefore, platinum has remained as the only viable metal

for use as the bottom electrode for the ferroelectric-dielectric film devices, having the

properties of high thermal conductivity, good stability in a high temperature oxygen

environment, and its high Schottky barrier height which gives rise to a low leakage

current.

Thus, the most commonly used metal electrode has been platinum. Since the adhesion

of Pt to SiO2 is poor, an adhesion layer has been typically used [21]. The material for

this adhesion layer is usually Ti. So the bottom electrode combination for most

ferroelectric materials is Pt/Ti/SiO2/Si.

A serious problem associated with the utilization of Pt/Ti/SiO2/Si electrode stack is the

formation of hillocks at the surface of the substrate at elevated annealing

temperatures. Hillocks are thin elevated features which protrude from the metal

surface (Pt in this case) during thermal cycling [21]. The formation of hillocks has

26

been an obstacle in the development of ferroelectric devices, because they may cause

the electrical short in the ferroelectric capacitor. The formation of hillocks in the

Pt/Ti electrodes has been reported by many researchers. The hillock is believed to be

related to compressive stress relief in Pt films which is mainly caused by the

oxidation of Ti diffused along the grain boundaries during annealing [22].

2.4. Chemical Solution Deposition Technique

2.4.1. Process Chemistry

Chemical solution deposition techniques namely sol-gel processing and metalorganic

deposition are extensively used for production of perovskite ceramic thin films and also

specifically for barium strontium titanate thin films.

The wet chemical methods such as sol-gel and metalorganic deposition (MOD) have

advantages over other thin film deposition techniques in that better composition control

and better homogeneity can more easily be achieved. In addition, the process may be

more economical for production, because no high vacuum systems are needed [23].

This lower capital investment cost also made this technique attractive not only for

making usable devices but also for scientific research.

Materials that are used in modern ceramic and device technology require high purity

and close control over composition and microstructure. Since the chemical reactants for

CSD can be purified conveniently by distillation and crystallization, films of high purity

can be fabricated by CSD [24]

A very important advantage of CSD is that as the elements that will form the final

compound mix at the molecular level, this means that the diffusion distances in the

inorganic film after pyrolysis required to achieve thermodynamically stable phases are

very short. This ultimate mixing and high reactivity also results in homogeneous and

dense films [24].

27

The basic principle of chemical solution deposition is to prepare a solution of elements

of the desired compound in a solvent and then coating this solution to a substrate and let

the solution polymerize to form a gel and then fire this gel to obtain an inorganic oxide

[24]

There are essentially two kinds of sol-gel technology. The first or colloidal method

involves the dispersion of colloidal particles in a liquid to form a sol and then the

destabilization of the sol to form a gel. The second method involves the

polymerization of organometallic compounds such as alkoxides to produce a gel with

a continuous network. The preparation of BST films in this thesis is based on

forming a solution of alkoxides of Ba, Sr and Ti.

Various precursors-solvents and stabilizers are employed in the production of films.

Since, the precursors (and solvents) used in the production determines the final

product, different precursors yield different solutions and films.

Important and typical precursors for making solid solutions are alkoxides of the general

composition M(O-R)n, where R is an alkyl radical (CH3, C2H5, etc.). Their properties

and reactions affect the preparation process and determine the product features.

The ideal compound to be used as precursors should satisfy following criteria:

• It should have high metal content: To minimize the volume change during the

change from metalorganic solution to inorganic film.

• High solubility in a common solvent with other starting compounds.

• It should be chemically compatible with other compounds in the formulation.

• Cost effective to produce: As the capital equipment requirements are small

(e.g. no high vacuum systems are needed), the cost of sol-gel is low, and the

precursors should not change this advantage.

• Thermally decompose without evaporating, melting or leaving a carbon deposit.

28

Criteria for solvents:

• Solvents should have high vaporization rates: they should evaporate as quickly

as possible. Vaporization depends on vapor pressure and also interaction

between solvent and solute.

• Solvent must be carefully selected in order to get a solution with high

concentration of necessary components, proper viscosity, and surface tension

[24].

After coating the solution on the substrate, gelation; the transition from a solution into a

solid starts by hydrolysis and condensation. The reactions are as follows:

M(OR)n + H2O → M(OR)n-1(OH) + ROH………………...(Hydrolysis) Eq. [2.14]

M(OR)n + M(OR)n-1(OH) → M2O(OR)2n-2 + ROH………(Condensation) Eq. [2.15]

M(OR)n-1(OH) + M(OR)n-1(OH) → M2O(OR)2n-2 + H2O…(Condensation) Eq. [2.16]

By hydrolysis and condensation of alkoxides, a polymeric product is formed,

organometallics and solvents reacts to form M-O-M (Metal-Oxide-Metal) bonds

leading to gelation of the solution. To some extent, it may be possible to control the

degree of gelation by using proper amount of water. Following parameters control

degree of gelation: Water amount, solvent, temperature, complex ligands, and pH value.

In film coating, the homogeneity of the gels is very important. The major advantage of

sol-gel processing for the preparation of ceramic materials is the fact that in a dried

homogeneous gel, the atoms are arranged in a manner closer to that in the desired final

crystalline phase than would be the case in a mixture of crystallites of the same

composition. Sintering behaviour is therefore improved [24].

29

2.4.2. Spin Coating Process

There are mainly two methods for the coating of solutions in CSD process: Dip coating

and spin coating. Spin coating is used more extensively than dip coating and it is the

process used in this study so only it will be described.

Bornside et al. [25] divide spin coating process into four stages: deposition, spin-up,

spin-off, and evaporation, although evaporation may accompany the other stages.

Figure 2.10 shows the stages of spin coating schematically.

1st stage 2nd Stage

3rd Stage 4th Stage

Figure 2.10. Four stages of spin-coating shown schematically [25].

30

An excess of liquid is dispensed on the surface during the deposition stage. For many

solutions it is beneficial to dispense through a sub-micron sized filter to eliminate

particles that could lead to flaws. It is important that the solution wets the surface of

the substrate completely. In the spin-up stage, at which the substrate is accelerated to

the desired rotation speed, the liquid flows radially outward, driven by the centrifugal

force. In the spin-off stage, excess liquid flows to the perimeter and leaves as

droplets. This stage is characterized by gradual fluid thinning. Fluid viscous forces

dominate the thinning behaviour. As the film thins, the rate of removal of excess

liquid by spin-off slows down, because the thinner the film, the greater the resistance

to flow, and because the concentration of the non-volatile components increases

raising the viscosity. In the fourth stage, evaporation takes over as the primary

mechanism of thinning. A spun film arrives its final thickness by evaporation after the

film becomes so thin and viscous that its flow stops [26].

Advantage of spin coating is that a film of liquid tends to become uniform in thickness

during spin-off and, once uniform, tends to remain so provided that the viscosity is no

shear dependent and does not vary over the substrate. This tendency arises due to the

balance between two main forces: centrifugal force, which drives flow radially outward,

and the viscous force (friction), which acts radially inward.

The thickness h(t) of an initially uniform film during spin-off is described by [27],

h(t) = h0/(1 + 4�w2h02t / 3�)1/2 Eq. [2.17]

where h0 is the initial thickness, t is time, and w is the angular velocity: solution density

� and solution viscosity � assumed constant. Even films that are not uniform initially

tend monotonically toward uniformity, sooner or later following the above equation.

Spinning is initially done at a slow rate to assure complete coverage of the substrate,

followed by fast spinning for a longer duration.

31

2.4.3. BST Solution Preparation

2.4.3.1. Solvents and Precursors

For BST solutions, mainly alkoxides are used as precursors. In Table 2.4, some

precursors and solvents used for BST solution preparation are listed.

For Ba and Sr precursors usually Ba and Sr acetates are preferred, whereas for

titanium, either titanium (IV) isopropoxide or titanium (IV) butoxide is preferred.

The molecular modification caused by chelation (modification of a chemical species

by reacting with a solvent) reduces the susceptibility of the starting reagents to

hydrolysis and condensation. This is important because the different hydrolysis and

condensation rates of different alkoxides cause formation of inhomogeneous gels

[24]. Titanium alkoxide is usually stabilized using chelating agents such as acetic

acid and acetylacetone. The type of the precursors, thus the precursor structure

influences the microstructure and dielectric properties. Usually, acetates are

dissolved in acetic acid, then Ti precursor which can be stabilized before is added to

this solution, and finally acetic acid or alcohol is added to dilute the solution.

2.4.3.2. Drying, Firing, Annealing of BST films

Thermal decomposition and crystallization behavior of the films is strongly dependent

on precursors. The solvents evaporate at 150˚-250˚C. The Ba and Sr precursors

decompose into carbonates at temperatures depending on the chain length of the

alkoxide precursors. For short alkyl chain length alkoxides, including acetates and

propionates, carbonates of Ba and Sr are formed at approximately 350˚C. For acetate-

based precursors, BaCO3 and SrCO3 are stable up to 550˚C, with TiO2 phase. Generally

for alkoxides with short-chain alkyl groups, independent of the titanium alkoxide used,

the crystallization of BST phase is delayed due to a formation of intermediate phase at

temperatures around 650˚C. This phase is said to be a complex carbonate phase,

(Ba,Sr)2Ti2O5CO3. Then above this temperature, carbonates react to form crystalline

(Ba,Sr)TiO3. The crystallization is complete at 700-800˚C [31].

32

Table 2.4. Different precursors and solvents for some sol-gel derived BST films.

Precursors Solvents Investigators

Barium acetate

Strontium Acetate

Titanium butoxide

Glacial acetic acid

Acetylacetone [14]

Barium acetate

Strontium Acetate

Titanium (IV) isopropoxide

Glacial acetic acid

Ethylene Glycol [13]

Barium acetate

Strontium Acetate

Titanium (IV) isopropoxide

n-propanol

Glacial acetic acid [28]

Barium carbonate

Strontium carbonate

Titanium (IV) isopropoxide

Ethylene glycol

Citric acid

[29]

Barium acetate

Strontium Acetate

Tetrabutyl titanate

Methanol

Glacial acetic acid

Ethylene glycol

dimethyl ether

Acetylacetone

(stabilizer)

[30]

33

For sol-gel derived barium titanate films, the decomposition pathway from the

amorphous pyrolyzed material into the desired microstructure is believed to take

place via intermediate carbonate (BaCO3 and TiO2) phases which first form at

temperatures between 300˚ and 400˚C. Other transitory phases, namely,

BaTi2O2CO5, have also been reported. Although BaTiO3 perovskite phase is

thermodynamically stable at temperatures as low as 410˚C, typically, it is not

observed in solution-deposited films until temperatures of 550˚C or 650˚C.

Evidently, the rate of decomposition of species such as BaTi2O2CO5 or the reaction

rate of titania and BaCO3 to yield BaTiO3 as shown below is not rapid at

temperatures between 400˚C and 600˚C:

BaCO3 + TiO2 → BaTiO3 + CO2 Eq. [2.18]

These results for the retarded formation of BaTiO3 are apparently in agreement with

a study of the crystallization rate of BaxSr1-xTiO3 thin films compared to PbTiO3 and

lead zirconate titanate (PZT). It was found that the crystallization rate in solution

derived BaxSr1-xTiO3 thin films was 300 times slower than in PbTiO3 [18].

34

CHAPTER 3

EXPERIMENTAL PROCEDURE

3.1. Solution Preparation

The BaxSr1-xTiO3 solutions with compositions in the ferroelectric (x 0.75) and

paraelectric (x<0.75) range were produced. The ferroelectric compositions were

Ba0.8Sr0.2TiO3 and Ba0.9Sr0.1TiO3 whereas the paraelectric compositions were

Ba0.5Sr0.5TiO3 and Ba0.7Sr0.1TiO3. The following table (Table 3.1) shows the

compositions produced in this study.

Table 3.1. BST solutions with different compositions and their corresponding

polarization states.

Composition Produced Polarization State

Ba0.9Sr0.1TiO3 Ferroelectric

Ba0.8Sr0.2TiO3 Ferroelectric

Ba0.7Sr0.3TiO3 Paraelectric

Ba0.5Sr0.5TiO3 Paraelectric

3.1.1. Starting Materials

For solution preparation two different routes were tried. So, both routes will be

explained. First solution was stable only for a week and the films produced by using

the first route cracked even for the highly diluted films, for film thickness greater

35

than 400 nm. That’s why a second solution system was tried. The precursors for the

first solution were Ba acetate, Ba(CH3CO2)2 [Aldrich-USA], Sr acetate,

Sr(CH3CO2)2 [Aldrich-USA], Ti(IV) isopropoxide, Ti(OCH(CH3)2)4 [Aldrich-USA],

and the solvents were glacial acetic acid, CH3COOH [Merck-Germany] and 2-

propanol (CH3)2CHOH [Merck-Germany]. In the second solution, 2-propanol was

substituted by ethylene glycol, HOCH2CH2OH [Merck-Germany].

3.1.2. Solution Preparation

The first solution preparation method was adapted from Dien et. al. [28]. First, Ba &

Sr acetates (for x = 0.5, equimolar amounts) were simultaneously dissolved in acetic

acid at room temperature, at acetic acid / acetates mole ratio 25:1 and stirred by

magnetic stirrer. A clear solution was obtained after an hour. Heating acetic acid to

80˚C accelerates the dissolution of acetates. Then Ti(IV) isopropoxide was dissolved

relatively easily in 2-propanol in another beaker by stirring (molar ratio 2-

propanol:Ti(IV) isopropoxide = 30:1). Then the two solutions were mixed to obtain

the parent solution. To remove residual particles, ultrasonic stirrer was used. The

solution was filtrated by a filter paper and then stored. A flow chart of film

preparation is given in figure 3.1. Amount of acetic acid is important because acetic

acid prevents the reaction of Ti(IV) isopropoxide with water, which was formed by

the reaction of 2-propanol with acetic acid:

CH3-COOH + HO-CH(CH3)2 = CH3-COO-CH(CH3)2 + H2O Eq. [3.1]

The parent solution was diluted by adding acetic acid to obtain crack free films.

Dilution with acetic acid was done at molar ratio of acetic acid / Ti(IV) isopropoxide

80:1. The diluted solution was not stable as the concentrated one. It was stable for 2-

3 days, whereas the concentrated solution was stable for a week.

The second solution preparation method (Figure 3.2) was adapted from Tahan et. al.

[32]. 25 milliliters of solution is taken as a base. Acetic acid to ethylene glycol ratio

36

Figure 3.1. Flow chart for solution preparation (first route) and coating of films.

Dissolve in (CH3)2CHOH (2-propanol)

and mix for 0.5 h

Ba(CH3CO2)2 (Ba acetate)

Dissolve in CH3COOH (acetic acid)

and mix for 1 h

Sr(CH3CO2)2 (Sr acetate)

Ti(OCH(CH3)2)4 Ti(IV)isopropoxide

Mix the solutions together for 0.5 h

Filter the solution

Spin coating 4000 rpm, 30s

Drying at 200˚C on hot plate for

10 min

Pyrolysis at 400˚C on hot

plate for 10 min

Sintering in furnace at 800˚C,

3 h

For obtaining desired thickness

37

Figure 3.2. Flow chart for solution preparation (second route) and coating.

Ba(CH3CO2)2 (Ba acetate)

Sr(CH3CO2)2

(Sr acetate)

Dissolve acetates in acetic acid

(CH3COOH) and stir by magnetic

stirrer for 1 h

Add Ti(OCH(CH3)2)4

(Ti(IV) isopropoxide and

stir for 0.5 h

Add etylene glycol and stir

for 0.5 h

Filtrate the solution

Spin coating 4000 rpm, 30 s

Drying at 200˚C on hot plate for

10 min

Pyrolysis at 400˚C on hot

plate for 10 min

Sintering in furnace at 800˚C

3 h

To obtain the desired thickness

38

was taken as 3:1, as suggested. 0.4M and 0.75M solutions were prepared. In each

case, calculations were repeated to determine the amounts of alkoxides.

Calculation for 0.4M Ba0.5Sr0.5TiO3 solutions is provided as an example. Molarity is

defined as number of moles of solute / number of liters of solution

0.4M = x moles of BST = x moles of Ti(IV) isopropoxide / 0.025 liters of solution

Then, x = 0.4*0.025 = 0.01 moles of Ti(IV) isopropoxide, then for Ba0.5Sr0.5TiO3

composition, number of moles of Ba acetate = x / 2 = 0,005 = number of moles of Sr

acetate.

Molecular weights of alkoxides are: Ba acetate: 255.43 gr/mol, Sr acetate: 205.71

gr/mol, Ti(IV)isopropoxide: 284.26gr/mol. So, molecular weights of precursors that

will be added to the 25 ml solution are: Ba acetate: 2.84 gr, Sr acetate: 1.03 gr and

Ti(IV)isopropoxide: 1.27gr.

The amounts of ethylene glycol and acetic acid were calculated for 25 ml. solution,

using the molar ratio acetic acid to ethylene glycol 3:1. The amount of acetic acid

was 19.82 gr. whereas ethylene glycol added was 6.83 gr. Molecular weights of these

solvents were: Acetic acid: 60.05 gr/mol, Ethylene glycol: 62.07 gr / mol; 1 liters of

ethylene glycol is 1.11 kg, whereas 1 liters of acetic acid is 1.05 kg.

First, Ba & Sr acetates (for x = 0.5, equimolar amounts) were simultaneously

dissolved in acetic acid at room temperature, and mixed by magnetic stirrer. All

acetic acid was added at once in order to prevent the formation of small aggregates of

acetates. A clear solution was obtained around an hour, exact dissolution time

depending on the concentration of the solution; the more acetic acid in the solution

the easier the dissolution of acetates. Then Ti(IV) isopropoxide was added to the

solution and stirred for 30 minutes. Finally, ethylene glycol was introduced to the

solution, and solution mixture was stirred for additional 30 minutes. It was observed

that ethylene glycol helps the dissolution of acetates. The solution was heated at

90˚C for 1 hour to promote the reaction between the components of the solution.

39

Then, the solution was cooled to room temperature, filtrated by a filter paper and

stored. The stock solution was stable for months.

Different alkoxides of different elements show a wide range of reactivity towards

water which makes the preparation of multicomponent homogeneous systems

difficult and also presents difficulties with premature gelation during film processing.

This can be overcome by adding a chelating organic ligand into the solution to

control the hydrolysis rates of the highly reactive alkoxide. Chelated solutions are

more stable in air and are easy to handle in processing [27].

When trying to incorporate titanium into a sol-gel solution, it is common to use one

of several commercially available alkoxides and to reduce its reactivity by modifying

it with some appropriate chemical modifier, due to the different activities of

alkoxides mentioned above. Acetic acid, as a modifier to the titanium alkoxide used

in our study, titanium isopropoxide (TIP), was studied by Birnie [33]. Acetic acid

(HOAc) is a popular modifier partly because it can easily dissolve a wide variety of

different precursor molecules, helping the creation of a multitude of multi-cation

solutions. However, the acetic acid can drive an esterification reaction with any

alcohol that is present, thus liberating water into the solution. The esterification

reaction can present some potential problems for sol-gel routes because, if the

reaction is not controlled, the water liberated can cause precursor condensation

reactions (and ultimately precipitation) in the solution. This is especially important

when TIP – HOAc solutions are heated in order to dissolve other precursor

components [33]. Thus, the gradual generation of water by the esterification reaction

(during subsequent processing) can put restrictions on the time and temperature that

might be used for carrying out these reactions. On the other hand, it has been pointed

out that the esterification reaction is an efficient method for introducing water

homogeneously into a solution and therefore avoiding the problems associated with

mixing irregularities. Since the water generation (and therefore the solution

hydrolysis) is a critical variable in sol-gel processing, it is important to understand

how this varies with time and temperature [33].

40

The alcohol used in this study is ethylene glycol and ethylene glycol reacts with

acetic acid to form an ester [34]:

HOCH2CH2OH + CH3COOH → HOCH2CH2OOCCH3 + H2O Eq. [3.2]

etylene glycol acetic acid ester

The water generated by this reaction enhances hydrolysis and condensation. The

addition of ethylene glycol as a cross-linking agent therefore leads to the

polymerization of the precursor solution resulting in stable M-O-M (M: Metal, O:

Oxide) linkages, with a decreased tendency to precipitation [32].

Coming back to the chelation, in the solution system used in this study, acetic acid

reacts with titanium(IV) isopropoxide, according to the equation:

M(OR)n + xCH3COOH → M(OR)n-x(OOCCH3)x + xROH Eq. [3.3]

The modification of alkoxide compounds dictates the structure and properties of

resulting species. The susceptibility of the starting reagents to hydrolysis and

condensation is reduced by chelation, along with changes in other precursor

properties such as pyrolysis behaviour [32].

Both solutions were diluted in order to prevent cracking and to decrease porosity.

3.2. Coating of Films on Substrates

3.2.1. Preparation of substrates

Prior to deposition, the surface of the substrates should be clean for good wetting and

adhesion of the film to the substrate. Si and Pt/Ti/SiO2/Si substrates were cleaned

using standard procedure. They were ultrasonically rinsed in acetone, followed by

rinsing in methanol and finally in deionized water and drying. Before cleaning, the

substrates were cut into small pieces out of a wafer. The dimensions of the pieces

41

were 1cm to 1cm, in square shape, as spin coating demands symmetrical substrates

for homogenous coating. The specifications of the Pt/Ti/SiO2/Si substrates

(INOSTEK - Korea) are given in Table 3.2.

Table 3.2. Specifications of Pt/Ti/SiO2/Si substrate (INOSTEK Korea).

SUBSTRATE: SILICON WAFER

Crystal Growth Method CZ (PRIME)

Orientation <100>

Thickness (�m) 525 ± 20

SiO2 LAYER

Film Growth Method Thermal Oxidation

Thickness (Å) 3000

Ti LAYER

Film Growth Method DC Magnetron Sputtering

Thickness (Å) 100

Pt LAYER

Film Growth Method DC Magnetron Sputtering

Thickness (Å) 1500

Orientation (111)

3.2.2. Spin Coating

A double-sided tape was used to stick the substrate on the head of the spin coater

then excess amount of solution was added on the substrate dropwise using Pasteur

pipettes. The solution was coated on substrates by spinning with a spin coater

(Chemat Technology Spin Coater) at 1000 rpm for 6 seconds for complete coverage

of the substrate, and then at different speeds for different durations. Optimum

spinning speed and time was determined as 4000 rpm for 30 seconds for 0.4M

solution.

42

Individual layers coated are thin. So, multiple coating – pyrolysis cycles were needed

to achieve the required thickness. The cracking of films due to shrinkage of films

during drying is the reason for making use of multiple coating of layers because there

is a limit for the thickness of the individual layer imposed by cracking

considerations. This limit is up to 1 �m generally and is dependent on the nature of

the precursors. For 0.4M solutions, each layer after coating and heat treatment was

200 nm thick, for spinning at 4000 rpm for 30 seconds.

3.2.3. Heat Treatment of the Films

Heat treatment step controls the microstructure of the films and microstructure

controls the properties of the material. Specifically for this study, electrical properties

like dielectric constant, dielectric loss and also ferroelectric properties depend on

heat treatment parameters: time and temperature. Appropriate heat treatment can

improve these properties by control of grain size and morphology; amount of

crystallization and by prevention of cracking.

The as-deposited films were amorphous. The films were sintered in order to obtain

crystalline perovskite films. The desired film thicknesses of 0.3 - 1 �m were obtained

by multiple coating – heat treatment cycles. Coated films were instantaneously

placed on a hot plate and dried at 200˚C for 10 minutes. After cooling to room

temperature, they were placed again on a hot plate preheated to 400˚C and left to

pyrolysis for 10 minutes.

Then, after obtaining the desired thickness by spin coating – pyrolysis cycles,

annealing (sintering) was done at 600˚, 700˚, 750˚ and 800˚C for 3 hours.

Temperature parameter was changed while keeping time parameter constant at 3

hours. The sintering was done in low temperature furnace under air atmosphere by

controlling the heating and cooling rates. The heating and cooling rate was 10˚C /

minutes in order to prevent cracking especially during cooling after the crystalline

BST film develops due to slight thermal expansion coefficient difference between

BST film and the substrate.

43

3.3. Characterization of Films

3.3.1. Structural and Morphological Analysis

The surface of the films was checked for cracks or inhomogeneities between the

cycles of coating, pyrolysis and sintering with optical microscope.

The crystal structure of the films was determined by x-ray diffraction using

Hanawalt’s method. Philips Diffractometer with monochromatic Co-K� radiation (�

= 1.79021 A˚) was used as diffractometer. 2� range between 20˚-80˚ was scanned.

Crystallinity and possible orientation of the film was also checked by investigating

the diffractograms and comparing with the powder diffractograms.

The surface and cross-section of the films were examined by scanning electron

microscopy (SEM-Jeol JSM-6400). BST films were coated by 250 Å nickel or gold as a

conductive layer to prevent discharge. Baltec SCD 050 sputter coater was used for Ni

coating. Thickness of the films was approximated from the cross-section SEM views.

The composition of the films could not be controlled with EDS (energy dispersive

spectroscopy) unit of the SEM. It can be seen from Table 3.3 that similar energy

values of barium L�1 and L�1 and the titanium K�1 and K�1 peaks (4.465, 4,827, 4.510

and 4.931 keV) impose difficulties on peak deconvolution.

It is therefore not possible to get quantative information using EDS. Also, another

problem arises due to the submicron thickness of the films. Due to this small

thickness electron beam easily penetrates through the film and accesses the

underlying electrode material stack. The contribution of x-rays from Ti layer present

on the substrate below Pt also deflects the total number of counts from Ti.

44

Table 3.3. X-ray energies of elements of BST film and Pt/Ti/SiO2/Si substrate.

Elements (Transitional energies in keV)

Ba Sr Ti O Pt Si

K�1 32.20 14.16 4,51 1,05 66,85 1,74

K�1 36,38 15,84 4,93 75,77 1,83

L�1 4,47 1,81 4,52 9,44

L�1 4,83 1,87 4,58 11,07

L�2 5,16 11,25

L�1 5,53 12,94

L�3 5,81 2,20 13,36

L� 5,95 1,58 0,40 8,27

M� 2,05

M� 2,13

M� 0,97 2,33

M� 0,60 1,60

3.3.2. Dielectric and Ferroelectric Measurements

Dielectric measurements were made by using HP 4194A Impedance-Gain Phase

Analyzer. MIM (Metal-Insulator-Metal) structure was formed in order to measure the

dielectric properties. The bottom metal electrode in MIM structure was Pt layer of

substrate (Pt/Ti/SiO2/Si – Inostek-Korea), Ti as an adhesion layer between Pt and SiO2.

BST film was served as the insulator. To form the top metal electrode, silver paint

(Agar Scientific) was used. The Ag solid compound was dissolved in IBMK

(isobutylmethylketone) and deposited as small points on BST layer. 4 points were

placed on each film, in order to see the possible change in measured properties. Figure

3.3 shows the Ag electrodes. Average radius of hemi-spherical like contacts vary

between 0.8-1 mm.

45

(a)

(b)

Figure 3.3. SEM micrographs showing the surface of electroded BST films, (a)

×400 (b) ×16.

46

Capacitance and dielectric loss of the films was measured in the frequency range 100

Hz -10 MHz. Capacitance of the films was converted to dielectric constant by using the

following formula assuming the parallel plate configuration:

dAK

C××

= 0ε Eq. [3.4]

where K is the relative dielectric constant, A is the area of the top electrode (Ag), �0

is the permittivity of free space (8.85*10-12 Farad/m), and d is the thickness of the

film.

The behavior of the films under voltage was tested by applying a dc voltage and

monitoring the corresponding capacitance change, again by using HP 4194A

Impedance-Gain Phase Analyzer.

Hysteresis loop was drawn by forming modified Sawyer-Tower circuit using a

waveform generator, a HP 54645A oscilloscope and a large capacitance capacitor.

The modified Sawyer-Tower circuit (Figure 3.4) was constructed by connecting a

large capacitance capacitor in series to the BST film. Since, the charge on the two

capacitors in series is the same, the voltage will drop mainly on the large capacitance

capacitor (capacitance = 10 nF) according to C = Q / V. The voltage applied (Vx) is

displayed on the x-axis on the oscilloscope, and voltage drop on the large

capacitance capacitor (Vy) is displayed on y-axis. The measurements were made at 1

kHz. Since, charges on both the sample and the large capacitor are the same (Qs =

Qy), the charge on the BST film (Qs) can be found by finding the charge on the 10 nF

capacitor which is equal to the multiplication of voltage drop on the large capacitor

(Vy) and its capacitance. The charge is converted to polarization, P = Qs/A, by simply

dividing the charge of the capacitor to electrode area and voltage applied is converted

to the electric field by dividing it to the film thickness, d. The equations followed in

the construction of P-E curve are given below:

P = Qs / A = (Cy * Vy) / A, Eq. [3.5]

47

E = Vx / d Eq. [3.6]

BST film, Cs, Qs

AC Signal Generator

Cy =10 nF

Figure 3.4. Schematic of the modified Sawyer-Tower circuit.

Vy

Vx

48

CHAPTER 4

RESULTS AND DISCUSSION

In this study, the aim was synthesizing BST solution by chemical deposition

technique and obtaining homogeneous and crack-free BST films with satisfactory

electrical properties. Two different systems of BST solution were tried in order to

obtain a homogenous solution and reduce cracking in the coated films. Processing

parameters like solution concentration, spinning speed, and sintering temperature

were varied to obtain well-crystallized, homogeneous and dense films. X-ray

diffraction and SEM were used for crystal structure determination and BST film-

substrate interface control, respectively. Once the optimum parameters were set,

effect of composition, sintering temperature and film thickness on electrical

properties was studied by measuring dielectric constant and loss. The change of

electrical properties with these parameters was correlated to microstructural changes.

Dielectric properties of the films were measured with changing frequency.

Capacitance – Voltage and ferroelectric measurements were also done.

Four different compositions of BST films were produced. Ba0.5Sr0.5TiO3 and

Ba0.8Sr0.2TiO3 were coated on Si substrates and Ba0.5Sr0.5TiO3, Ba0.7Sr0.3TiO3 and

Ba0.9Sr0.1TiO3 were coated on Pt/Ti/SiO2/Si substrates. Due to the reaction of BST

film and the Si substrate and resulting silicide and SiO2 formation, the electrical

properties of BST films that were coated on bare Si substrates were not measured;

they were utilized for crystallization studies only. Therefore, dielectric and

ferroelectric properties of BST films coated on Pt/Ti/SiO2/Si substrates of

composition Ba0.5Sr0.5TiO3, Ba0.7Sr0.3TiO3 and Ba0.9Sr0.1TiO3 were measured.

49

Table 4.1. Compositions of BaxSr1-xTiO3 films that were produced in this study

listed according to the substrates they were coated on.

Composition of the films Substrate Electrical Measurements

Ba0.5Sr0.5TiO3 (paraelectric) Si No

Ba0.8Sr0.2TiO3 (ferroelectric) Si No

Ba0.5Sr0.5TiO3 (paraelectric) Pt/Ti/SiO2/Si Yes

Ba0.7Sr0.3TiO3 (paraelectric) Pt/Ti/SiO2/Si Yes

Ba0.9Sr0.1TiO3 (ferroelectric) Pt/Ti/SiO2/Si Yes

4.1. Improvement of Film Quality

In the first solution preparation route, second solvent was 2-propanol in addition to

acetic acid. Films that were coated using that solution were cracked after 400 nm

even for the 0.28M solution. The reason of cracking for sol-gel derived films is the

large volume change during drying. During drying, wet solution changes into an

amorphous mixture of carbonaceous metal ions. The cracks during deposition of the

films are most likely to occur in the drying stage. In this stage, wet coating is

transformed into harder amorphous solid, during this transformation, considerable

shrinkage occurs, leading to cracks. When the solvent evaporates, a film containing

the desired elements forms on the substrate. Since the network film is a plastic or

viscoelastic solid, its tendency to crack during drying and firing is substantially

reduced [27].

Ethylene glycol is known to possess high boiling point and high latent heat of

vaporization, due to this property, ethylene glycol retains the atomic mobility and

reduces the tendency to crack [27]. Instead of 2-propanol, ethylene glycol was added

to the solution in order to prevent cracking of the films, following the same solution

route with Tahan et. al. [32]. By this addition, crack-free films up to 600 nm

thickness were obtained. In Figure 4.1, surface SEM micrographs of two films coated

50

by using first and second solutions are given. Surface cracks can be seen for the film

coated using first solution.

Figure 4.1. SEM micrographs, (top) showing surface cracks of 3 layers

Ba0.5Sr0.5TiO3 film (500 nm) coated by using first solution: solution containing 2-

propanol, (bottom) showing crack-free surface of 3 layers Ba0.5Sr0.5TiO3 film

(600nm) coated by using second solution: solution containing ethylene glycol, both

films were sintered at 800˚C for 3 hours.

51

Films may also crack due to thermal expansion coefficient differences between the

film (BST) and Pt layer of the substrate. Lopez et al. [34] relates the cracking to

mismatch in thermal expansion coefficients of BST and underlying Pt substrate.

Table 4.2 shows the thermal expansion coefficients and lattice parameters for bottom

Pt layer, BaTiO3 (BT) and SrTiO3.

Table 4.2. Thermal expansion coefficients of Pt, BaTiO3 and SrTiO3 [35, 53].

Material Thermal Expansion

Coefficient (1/˚C) Lattice Parameter (Å)

Pt 9×10-6 3.97

BaTiO3 9.8×10-6 3.994

SrTiO3 11×10-6 3,901

Although Lopez et al. had approximated thermal expansion coefficient of BST to that

of BT (he had taken it as 4×10-6 1/˚C), instead of averaging the thermal expansion

coefficients of BT and ST (which are 9.8×10-6 1/˚C and 11×10-6 1/˚C, respectively),

so that they assumed a larger difference in thermal expansion coefficients than it

really is, it is nevertheless probable that films might crack due to this difference. This

thermal expansion coefficient difference might develop strains especially during

cooling, after the formation of inorganic oxide [27]. That’s why controlled heating

and cooling was applied to the films, with heating and cooling rates as low as

10˚C/min.

Solution viscosity is an important parameter in sol-gel processing, which depends on

concentration, type of the solvent, and possible reactions between the components of

the solution. It can also be adjusted by controlled hydrolysis and by adding cross-

linking agents or by adding acid to destroy large molecules formed in the solution

[27]. Since film thickness and uniformity are dependent on solution viscosity, the

52

proper spinning speed also depends on viscosity. The films were coated by spinning

at different spinning rates in order to obtain homogeneous and uniform films. The

films were spun at 3000, 4000 and 7500 rpm (revolutions per minute) for 30 seconds

each. The surface of the films spun at 7500 rpm had a dull appearance, showing

regions of inhomogeneity whereas the films spun at 4000 & 3000 rpm had bright and

clear surfaces. Therefore, 7500 rpm was not an appropriate spinning speed. Coating

thickness is inversely proportional to spinning speed [26], so the higher spinning

speed results in lower thickness. So, in order to obtain a thinner layer, 4000 rpm was

chosen as the spinning speed. The thickness of the individual layer coated from 0.4

M solution, spun at 4000 rpm for 30 seconds was already high, around 200 nm.

4.2. Dilution of the Solution

BST solutions were diluted for three main purposes:

1. Layers coated from diluted solutions are thinner than those coated by using

concentrated solutions. So, during drying the sudden volume change occurring due to

evaporation of solvents is less for layers coated from diluted solutions than those

coated from concentrated ones. This prevents cracking due to sudden shrinkage. In

this study, the films derived from the second route are crack-free. However, as

explained before, that was the addition of ethylene glycol instead of 2-propanol not

the concentration of the solution which made the difference. Among the films

derived from the second route though, the ones derived from the 0.4 M solution are

crack-free compared to the ones derived from the 0.75 M solution.

2. Results indicate that, the porosity in the film increases with increase in precursor

solution concentration. The lower porosity in the film deposited using dilute solution

is due to the fact that individual layers are thin and hence thermally cycled more

times which results in effective pyrolysing of the organics compared to the film

prepared from the concentrated solution. In sol-gel preparation, the individual layer

on pyrolysis leaves behind pores in the film, which will be covered by subsequent

layers. The more the number of layers, the less the density of pores. Also, the

53

individual layers of the films prepared from the concentrated solution are thicker

than the film prepared from dilute solution, hence require less number of layers to

attain required thickness and incorporates more carbonatious decomposition products

in the film. On annealing to high crystallization temperature, these embedded

impurities burn-off leaving behind pores in the film and in the process changes the

microstructure [19].

3. By preparing thinner layers using the highly diluted solution, multiple nucleation

sites within the film are avoided, and the nucleation on the underlying layer (either

the substrate or the prior-crystallized Ba0.8Sr0.2TiO3 layer) can be enforced, which

results in layer-by-layer homoepitaxial growth within an individual grain [36].

However, that is the case when the solution concentration is very low (around

0.05M) and when rapid thermal annealing is applied to every layer produced. Due to

precursors and solvents we used, it was not possible to get such dilute solutions.

4.3. Crystalline Film Formation

The crystal structure of BaxSr1-xTiO3 thin films was determined by x-ray diffraction.

The analysis of diffractograms was made by Hanawalt method. Initially, only films

of composition Ba0.8Sr0.2TiO3 and Ba0.5Sr0.5TiO3 were coated on Si substrates to

study film formation. The effect of sintering temperature on the degree of

crystallization was also studied. The composition used for this comparative study

was Ba0.5Sr0.5TiO3. Three Ba0.5Sr0.5TiO3 films were coated using the same number of

layers and sintered at 600˚C, 700˚C and 800˚C. All films crystallized into cubic BST

phase, BST peaks completely fitting to the JCPDS file of Ba0.5Sr0.5TiO3, 39-1395.

The intensity of BST peaks sintered at 600˚C was quite large, showing well

crystallization of the films. The peaks were larger at 700˚C indicating a better

crystallinity was achieved. At 800˚C, the peaks were even larger, and the major [110]

peak was achieved to its maximum. The optimum crystallization temperature was

chosen as 800˚C, after comparing the diffraction peaks of films annealed at 600˚C,

700˚C and 800˚C. The x-ray diffraction spectra can be seen in Figure 4.2.

54

The composition difference between Ba0.5Sr0.5TiO3 and Ba0.8Sr0.2TiO3 films in terms

of diffraction angle, 2�, can be seen comparatively in Figure 4.3. Both films were

sintered at 800˚C for 3 hours. The d-spacings between the planes increase while 2�

values decrease with the decrease in the Sr content according to Bragg’s law (� =

2dsin(�)).

Figure 4.2. XRD spectra of Ba0.5Sr0.5TiO3 films on Si substrate sintered at 600˚C,

700˚C and 800˚C for 3 hours.

55

X-ray diffractograms can also be used as evidence for the differentiation and

stoichiometry of the composition of BST films. The complete fit of intensity and

position of diffraction peaks for Ba0.5Sr0.5TiO3 films to JCPDS file shows that the

stoichiometry holds, whereas the shift in the 2� values to lower angles for

Ba0.8Sr0.2TiO3 films indicates that the films contain less Sr. For Ba0.8Sr0.2TiO3 thin

films, JCPDS file of Ba0.77Sr0.23TiO3 was used (JCPDS: 44-0093).

Figure 4.3. XRD spectra of (a) 5 layers, 1 �m thick Ba0.5Sr0.5TiO3 film on Si,

sintered at 800˚C (b) 5 layers, 1 �m thick Ba0.8Sr0.2TiO3 film on Si, sintered at

800˚C.

56

There was no intermediate phase observed in the diffractograms. The peaks in the

20˚ - 80˚ range matched completely in intensity and angle with JCPDS files for both

Ba0.5Sr0.5TiO3 and Ba0.8Sr0.2TiO3 films. Some researchers have detected an

intermediate complex carbonate phase for BT and ST, and they claim that same

phase should be seen for BST, which is a solid solution of BT and ST. This

carbonate phase was reported only for the films derived from acetate-based

precursors (in general, long-chain carboxylate precursors) and below the

crystallization temperature of BST phase around 600˚C; up to 750˚C, independent of

Ti precursor used [30, 31, 37]. The exact stiochiometry of the intermediate phase is

not known, but Gopalakrishnamurthy suggested a composition Ba2Ti2O5CO3 for BT,

and a similar composition was suggested for BST: (Ba,Sr)2Ti2O5CO3; a different

suggestion was (Ba,Sr)TiO2CO3 [23]. Although there is no intermediate phase

detected for BST films coated on Si substrates at different sintering temperatures, it

is possible that films contain an intermediate phase. The intensity of intermediate

phase peaks were reported as low so the films might not be thick enough so that

reflections from the intermediate layer are less than required to differentiate the small

intermediate phase peaks in the diffractograms. As the thickness of the film

decreases, the number of counts taken from the film decreases relative to the counts

from the substrate so intensity of BST peaks decreases with decreasing thickness.

The effect of the intermediate phase on the crystallization behaviour was studied by

some researchers. Hasenkox et al. studied the effect of different precursors on the

crystallization behaviour [31]. They claim that amorphous BST film first crystallizes

into an intermediate phase before crystallizing into the perovskite structure.

According to them, this intermediate phase favors grain growth in the BST films

derived from acetate precursors compared to films prepared by using other

precursors than acetates. Because, the formation of some intermediate phase in the

films from acetate precursors means that the crystallization process still proceeds at

high temperatures where it has already terminated for thin films prepared by using

different precursors. Such a delayed crystallization favoring the growth of grains

combined with high mobility of the ions due to the high temperatures may also

support the densification of the film. These two factors can lead to the observed

57

increase in the permittivity of thin films prepared form the acetate precursor. They

also claim that this delayed crystallization can also be the origin of columnar grain

growth of ST films on platinized silicon substrates derived from acetate precursors.

This can be a point that justifies our choice of acetate precursors in this study.

For measuring the electrical properties, BST films of three different compositions

were coated on Pt/Ti/SiO2/Si instead of bare Si substrates. Since the adhesion and

growth of films on different substrates might be different, crystallization studies were

repeated for the films coated on Pt/Ti/SiO2/Si. Again, Ba0.5Sr0.5TiO3 composition

was used for this study. Figure 4.4 shows the x-ray diffractograms for Ba0.5Sr0.5TiO3

thin films coated on Pt/Ti/SiO2/Si substrates and sintered at 600˚C, 700˚C and 800˚C.

This time, BST film sintered at 600˚C showed peaks other than BST and substrate

(Pt) peaks. These peaks are suspected to be intermediate phase peaks, the one at

50.7˚ is determined to be a BaCO3 (JCPDS: 74-1628) peak, the other one at 28.8˚ is

Ba2Ti5O12 (JCPDS: 17-661). At 700˚C, these unfamiliar peaks of possible

intermediate phase disappeared but the intensity of the other peaks did not

significantly change. At 800˚C, intensity of BST peaks increased remarkably,

showing better crystallization of the films.

The above study was done for the films coated by using first BST route. Upon

addition of ethylene glycol, solution precursor chemistry thus thermal decomposition

and crystallization behaviour of the films changed, so this crystallization study was

repeated for the films coated by the second solution route on INOSTEK

Pt/Ti/SiO2/Si substrates. Figure 4.5 shows the x-ray diffractograms for Ba0.5Sr0.5TiO3

thin films coated on these Pt/Ti/SiO2/Si substrates and sintered at 700˚C, 750˚C and

800˚C using second solution. The film sintered at 750˚C had similar peak intensities

to film sintered at 800˚C, but intensity of peaks of BST film sintered at 700˚C was

lower than the other two.

58

Figure 4.4. X-ray diffractograms for Ba0.5Sr0.5TiO3 thin films coated on

Pt/Ti/SiO2/Si substrates and sintered at 600˚C, 700˚C and 800˚C using first solution

route.

59

The phase formation behaviour of Ba0.7Sr0.3TiO3 and Ba0.9Sr0.1TiO3 films was also

checked by x-ray diffraction. Figure 4.6 shows the x-ray diffractograms of

Ba0.9Sr0.1TiO3 (JCPDS card of BaTiO3 was used: 5-626), Ba0.7Sr0.3TiO3 (JCPDS card

of Ba0.67Sr0.33TiO3 was used: 89-0274) and Ba0.5Sr0.5TiO3 (JCPDS: 39-1395) films all

sintered at 800˚C for 3 hours. All films showed well crystallization with intense

peaks. In order to differentiate between the different films better, the (110) peaks of

the films are given in figure 4.7. The shift in the diffraction angle of the peaks to

lower degrees with the increase in Sr content of the films is obvious. This shift was

in accordance with JCPDS cards of the films, suggesting that the composition of the

films were correct.

s

Figure 4.5. X-ray diffractograms for Ba0.5Sr0.5TiO3 thin films coated on different

Pt/Ti/SiO2/Si substrates and sintered at 700˚C, 750˚C and 800˚C using second

solution route (using ethylene glycol).

60

Figure 4.6. X-ray diffractograms of Ba0.5Sr0.5TiO3, Ba0.7Sr0.3TiO3 and Ba0.9Sr0.1TiO3

films (from bottom to top) sintered at 800˚C for 3 hours.

61

Figure 4.7. X-ray diffractograms of (110) peaks of Ba0.5Sr0.5TiO3, Ba0.7Sr0.3TiO3 and

Ba0.9Sr0.1TiO3 films (from bottom to top) sintered at 800˚C for 3 hours.

62

4.4. Morphology of the BST Films

Surface of the films was examined by optical microscopy and SEM for possible

cracks and inhomogeneities. The films prepared from 0.4M solution (second solution

containing ethylene glycol) had crack-free and homogeneous surface up to 600 nm

thickness (Figure 4.1(b)). Since the grains of BST thin films synthesized by sol-gel

processing are typically in the order of 20-70 nm. [13, 29, 38] excluding epitaxially

grown films [14, 31], magnification of SEM is not enough to see individual grains and

grain morphology. In this study, average grain size and grain size distribution couldn’t

be determined by using SEM for this reason. But, average grain sizes were estimated by

using Scherrer’s Formula, which is a technique based on measuring the full width of x-

ray diffraction peaks at the half maximum height of the peak.

The formula is as follows [39]:

θλ

cos94.0×

×=B

t Eq. [4.1].

where, t is the average grain size, � is the wavelength of radiation of the x-ray beam

used, B is the width of the peak at the half of the maximum intensity (in radians) and �

is the half of the diffraction angle 2�. The contribution to the peak broadening from

instrumental broadening must be substracted by the help of a reference sample,

however, since instrumental contribution is only significant for large particle sizes (t>30

nm), it was neglected. For BST grain size determination, the most intense diffraction

peak, (110) was used. Among the three Ba0.5Sr0.5TiO3 films in Table 4.3, the one

sintered at 600˚C for 3 hours had an average grain size of 16.9 nm. The average grain

size didn’t change much for the one sintered at 700˚C but increased to 25.4 nm for the

film sintered at 800˚C for 3 hours (Table 4.3). To increase the accuracy in the width

measurements, (110) peak was studied separately by making diffraction studies using

1/8˚ per second speed and three different other diffractograms corresponding to

Ba0.9Sr0.1TiO3, Ba0.7Sr0.3TiO3, and Ba0.5Sr0.5TiO3 were obtained (Figure 4.7). The grain

size did not change much with Sr content, it was 30 nm approximately.

63

Table 4.3 Grain sizes of 1 �m thick BST films sintered at different temperatures.

Composition Sintering Temperature Grain Size (nm) Ba0.5Sr0.5TiO3 600˚C 16.9

Ba0.5Sr0.5TiO3 700˚C 17.2

Ba0.5Sr0.5TiO3 800˚C 25.4

Ba0.5Sr0.5TiO3 800˚C 33.9

Ba0.7Sr0.3TiO3 800˚C 27.5

Ba0.9Sr0.1TiO3 800˚C 29.6

The cross-section of BST films was examined by SEM for determining film

thickness and checking the quality of Pt / BST interface. The thickness of the films

changed from 400 nm to 1 �m accordingly with the number of layers coated. For

example, 0.4M solution derived film spun at 4000 rpm for 30 seconds, and coated by

2 coating-pyrolysis cycles was 400 nm, whereas another film coated by 3 cycles was

600 nm thick. SEM micrographs of these films are shown in Figure 4.8 and 4.9,

respectively. The interfaces between the BST and Pt were distinct and homogeneous.

As mentioned in Chapter 3, the composition of BST thin films cannot be checked by

energy dispersive analysis of SEM, due to overlapping of x-ray peaks of elements

constituting the film. Therefore, we made use of EDS in order to understand whether

any impurities exist in the film. Analysis indicated that the elements of BST: Ba, Sr,

Ti, and O are found in the films without any residuals, e.g. carbon. The cross-section

of a Ba0.5Sr0.5TiO3 film obtained from the first route coated on Si substrate is shown

in the Figure 4.10. The EDS figures of BST film and the substrate are included in

Figure 4.11 and 4.12 respectively. The overlapping of energies of elements Ba, Ti

and Si, Sr can be seen in the EDS Spectra. Nickel (Ni) came from conductive coating

for preparing the sample to SEM analysis.

64

Figure 4.8. Cross-section SEM micrograph of 2 layers Ba0.5Sr0.5TiO3 film on

Pt/Ti/SiO2/Si substrate, (top ×5000, middle ×10000, bottom ×20000).

65

Figure 4.9. Cross-section SEM micrograph of 3 layers Ba0.5Sr0.5TiO3 film on

Pt/Ti/SiO2/Si substrate, sintered at 800˚C for 3 hours.�

�

�

�

�

�

�

�

�

�

�

Figure 4.10. SEM micrograph of Ba0.5Sr0.5TiO3 film on Si substrate.

66

Figure 4.11. EDS spectra of Ba0.5Sr0.5TiO3 film on Si substrate, sintered at 800˚C for 3 hours.

Figure 4.12. EDS spectra of Si substrate.

Same elemental analysis was repeated for BST films coated on Pt/Ti/SiO2/Si

substrates. EDS spectra of the 6 layers Ba0.5Sr0.5TiO3 film and underlying

Pt/Ti/SiO2/Si substrate is shown in Figure 4.13 and 4.14, respectively. Presence of

67

Au in the EDS spectras is due to conductive coating layer for preventing discharge

during SEM analysis.

Figure 4.13. EDS spectra of�6 layers Ba0.5Sr0.5TiO3 film on Pt/Ti/SiO2/Si substrate.

Figure 4.14. EDS spectra of Pt/Ti/SiO2/Si substrate.

68

4.5. Dielectric Properties of BST Films

Dielectric properties of films like dielectric constant and loss depends on

microstructural features of the films like grain size, crystallinity, porosity and hence

give valuable information about the film’s quality. In this study, dielectric constant

and loss of the films that differ in composition, thickness and heat treatment schedule

were measured.

Capacitance and dielectric loss of the BST films were measured by HP 4194A

Impedance-Gain Phase Analyzer. These properties were measured in the frequency

range 100 Hz - 1 MHz, with an oscillating AC voltage of 0.5 V. There was some

dispersion in the capacitance of films with frequency. The change of capacitance and

dielectric constant calculated from those capacitance values of 600 nm Ba0.7Sr0.3TiO3

film with frequency are given in Figures 4.15 and Figure 4.16, respectively.

Capacitance versus Frequency

1

1,5

2

2,5

3

3,5

1 10 100 1000

Frequency (kHz)

Cap

acita

nce

(nF)

Figure 4.15. The change of dielectric constant with frequency for 3 layers

Ba0.7Sr0.3TiO3 film (t = 600 nm).

69

Dielectric Constant vs frequency

150

180

210

240

270

300

330

1 10 100 1000

Frequency (kHz)

Die

lect

ric

Con

stan

t

Figure 4.16. The change of dielectric constant with frequency for 3 layers

Ba0.7Sr0.3TiO3 film (t = 600 nm).

In order to see the effect of degree of crystallization on dielectric properties,

dielectric constant of films having the same thickness, sintered at different sintering

temperatures was measured. Three Ba0.5Sr0.5TiO3 films, 3 layers in thickness, 600

nm, were sintered at 600˚C, 700˚C and 800˚C. There was an increase in dielectric

constant (measured at 1 kHz) with increasing sintering temperature (Figure 4.16).

This increase is due to better crystallinity of the films and increase in grain size at

higher sintering temperatures. Also, film porosity is reduced at high sintering

temperatures. As previously given x-ray diffraction results also showed, at 600˚C and

700˚C, the films were poorly crystallized compared to the films sintered at 800˚C.

This was confirmed by lower dielectric constant of these films with respect to the

film sintered at 800˚C.

70

Dielectric constant versus sintering temperature

0

50

100

150

200

250

500 600 700 800 900

Sintering temperature (C)

Die

lect

ric

Con

stan

t

Figure 4.17. Change in dielectric constant (at 1 kHz) with different sintering

temperatures, for 3 layers Ba0.5Sr0.5TiO3 films.

The dielectric properties of BaxSr1-xTiO3 films depend on Ba/Sr ratio and these

composition tunable properties were investigated by monitoring the change in

dielectric properties with composition. For this study, Ba rich portion of continuous

solid solution scale was chosen. 2 layers and 3 layers Ba0.5Sr0.5TiO3, Ba0.7Sr0.3TiO3

and Ba0.9Sr0.1TiO3 films were produced and their dielectric properties were

compared. The dielectric constant of 3 layers BST films with different compositions

as a function of frequency are shown in Figure 4.18. The highest dielectric constant

was measured for 3 layers Ba0.7Sr0.3TiO3 film (approximately 600 nm in thickness)

and it was 313 at 1 kHz. The other two films that were 3 layers thick had lower

dielectric constants, Ba0.5Sr0.5TiO3 film had a dielectric constant of 230 and

Ba0.9Sr0.1TiO3 film had 228. Other researchers also observed similar trend in the

change of dielectric constant of BST films with composition. For example, Tahan et

al. also obtained highest dielectric constant for Ba0.75Sr0.25TiO3 films at room

temperature; for 400 nm film, they measured 440 at 1 kHz [13]. Also, Pontes et al.

reported higher dielectric constant for Ba0.8Sr0.2TiO3 film than Ba0.4Sr0.6TiO3 film,

749 and 680 respectively at 1 kHz [40]. It is known that Curie temperature of bulk

71

BST decreases with increasing strontium content according to the approximation: Tc

= 371x – 241, where x is the fraction of barium [11]. So, addition of 0.3 mole percent

Sr would bring the Curie temperature of the BaTiO3 to room temperature, where the

dielectric constant is maximum (Figure 4.19). So, this explains why Ba0.7Sr0.3TiO3

film displays the highest dielectric constant and why Ba0.5Sr0.5TiO3 and

Ba0.9Sr0.1TiO3 films have similar dielectric constants. The latter two compositions are

equally distant from the transition temperature where the dielectric constant peaks at

room temperature.

Dielectic Constant vs frequency, with composition

150

190

230

270

310

350

1 10 100 1000

Frequency (kHz)

Die

lect

ric

Con

stan

t

3 layers, 0.5-0.53 layers, 0,7-0,33 layers, 0,9-0,1

Figure 4.18. Dielectric constant of BST films with different compositions as a

function of frequency for 600 nm films.

Similar composition dependence behaviour was also observed for 2 layers

(approximately 400 nm) BST films. Again Ba0.7Sr0.3TiO3 film has shown the highest

dielectric constant, 176 at 1 kHz measuring frequency, but this time dielectric

constant of Ba0.9Sr0.1TiO3 film was close to that of Ba0.7Sr0.3TiO3, it was 175.

Ba0.5Sr0.5TiO3 had the lowest of three, 119 at 1 kHz. The dielectric constant of 2

72

layers BST films with different compositions as a function of frequency are shown in

Figure 4.20.

Figure 4.19. The decrease in the Curie temperature of BaTiO3 with Sr addition [11].

Dielectric Constant vs frequency, with composition

60

80

100

120

140

160

180

200

1 10 100 1000

Frequency (kHz)

Die

lect

ric

Con

stan

t

2 layers0.5-0.52 layers0.7-0.32 layers0.9-0.1

Figure 4.20. Dielectric constant of BST films with different compositions as a

function of frequency for 400 nm films.

73

Dielectric properties of some sol-gel derived BST films from literature varying in

composition, thickness, precursors and heat treatment temperatures are given in

Table 4.4 for comparison. It can be seen that dielectric constant of BST films

obtained in this study are comparable with other sol-gel deposited films but lower

than that of epitaxial films. This may be due to the variation in the microstructure. In

the epitaxial films crystallization is highly oriented, normally vertical to the surface

of the films, and this provides a stronger polarization direction. This highly oriented

structure tends to form a concentrated polarization which results in higher dielectric

constant. But in the present case the films were polycrystalline with random

orientation, which implies an almost equal polarization in all directions [41].

Film stoichiometry should also have a pronounced effect on dielectric constant.

Several groups have pointed out that the permittivity at room temperature is

maximum when the (Ba+Sr)/Ti is 1:1, and decreases when the films are either

titanium rich or titanium poor [42, 43]. Similar results have been obtained for

SrTiO3. The dependence on (Ba+Sr)/Ti ratio becomes much less marked when the

film thickness is decreased [42]. Since, EDS analysis couldn’t be used for BST films

due to overlapping of x-ray energy peaks of Si-Sr and Ba-Ti, the exact stoichiometry

of the films couldn’t be checked. Therefore lower dielectric properties compared to

other sol-gel derived films, encountered in the present case may be due to possible

non-stoichiometry in the films.

Dielectric loss of the films was also measured. For all BST films, independent of

composition, thickness and sintering temperature, dielectric loss showed the same

dispersion characteristics, it was nearly constant with frequency in the 1 kHz-100

kHz range, then it increased in 100 kHz-1 MHz range. Same dispersion

characteristics were also observed for other BST films [38, 40]. In figure 4.21, the

dielectric loss with respect to frequency for 3 layers of BST films with different

compositions is shown. Dielectric loss of the films was between 4 and 7 % for

Ba0.5Sr0.5TiO3 and Ba0.9Sr0.1TiO3 films up to 1 MHz, whereas it changed from 7 to 10

% for Ba0.7Sr0.3TiO3. There is a tendency for the dielectric loss to increase

remarkably above 1 MHz for all films. Several possible sources exist for such

74

Table 4.4. Dielectric properties of some sol-gel derived films with different

processing parameters. Properties of films that were deposited using other common

BST deposition techniques are also provided to give an idea.

Reference

Thickness

Heat

treatment

Composition

Dielectric Loss (%)

Dielectric constant

[13].

400 nm

700ºC, 1h

Ba0.75Sr0.25TiO3

4 440 (@1kHz)

[40]

400 nm

700ºC, 2h

Ba0.4Sr0,6TiO3

Ba0.8Sr0.2TiO3

1 4

680 (@100

kHz) 749 (@100

kHz)

[14]

300 nm

750ºC, 10 min (RTA)

Ba0.8Sr0.2TiO3

Not reported

680 (@100 kHz)

[41]

400 nm

700˚C, 2h

Ba0.7Sr0.3TiO3

2,36

120

(@100kHz)

[28]

600 nm

800ºC, 30

min Ba0.67Sr0.33TiO3

Not reported

350 (@1kHz)

[38]

160 nm

800˚C 1 min (RTA)

Ba0.5Sr0.5TiO3

2

230

(@100kHz)

[44] (PLD) 40 nm 700˚C, 1h Ba0.5Sr0.5TiO3 3,5 150 (@1 MHz)

[45] (MOCVD) 30 nm 625˚C Ba0.7Sr0.3TiO3

Not reported

208 (@1 kHz)

[46] (RF-Sputtering) 80 nm 650˚C Ba0.5Sr0.5TiO3 2 530 (@10

kHz)

75

dispersion, including the finite resistance of the electrodes, the presence of a barrier

layer between the insulating film and the electrode surface, or leaky grain boundaries

[40].

Dielectric Loss vs frequency, with compositon

0

0,050,1

0,150,2

0,250,3

0,35

1 10 100 1000

Frequency (kHz)

Die

lect

ric

Loss

600 nm, 0,5-0,5600 nm, 0,7-0,3600 nm, 0,9-0,1

Figure 4.21. Dielectric loss as a function of frequency for 600 nm (3 layers) BaxSr1-

xTiO3 films.

Dielectric loss originates from two mechanisms: resistive loss and relaxation loss.

Resistive loss mechanism involves energy consumption by the mobile charges in the

film; whereas, in the case of relaxation loss mechanism, it is the relaxation of dipole

which dissipates the energy. If there are very few charges in the film then the latter

mechanism is dominating. The resistive loss mechanism is directly connected to the

leakage current of the film: if the leakage current is higher the loss is also higher.

Alternatively, if the dielectric constant of the film is large, then an increase in

dielectric loss is obvious due to the contribution from the relaxation mechanism.

Enhanced polarization increases the dissipation during the relaxation [19]. In our

films, dielectric loss at 800˚C is lower than the value at 600 and 700˚C (Figure 4.22).

This may be due to achievement of lowest resistive loss due to complete

crystallization into pure BST phase. Although relaxation component of dielectric loss

is higher at 800˚C, resistive loss is lower so overall dielectric loss is lower than that

76

of 600 and 700˚C. However, this can only be a possible explanation. In order to

verify this suggestion, leakage current measurements should be made. Since,

resistive loss component of dielectric loss is directly proportional to leakage current,

it may be possible to estimate the magnitude of resistive loss of the films sintered at

different temperatures.

Dielectric loss vs frequency, with temperature

0,02

0,04

0,06

0,08

0,10

1 10 100 1000

frequency (kHz)

Die

lect

ric

loss

700600800

Figure 4.22. Dielectric loss of Ba0.5Sr0.5TiO3 films as a function of frequency

sintered at different temperatures for 600 nm films.

The above paragraph explains the causes of the high dielectric loss for Ba0.7Sr0.3TiO3

film evident from Figure 4.20. Since, dielectric constant for this composition is the

highest amongst the others, dipole moment density is also highest, resulting in an

increase in dipole relaxation loss. As the dielectric constants thus dipole moments

densities of Ba0.5Sr0.5TiO3 and Ba0.9Sr0.1TiO3 films are comparable, their dielectric

loss values are also comparable.

77

The dependence of dielectric constant and loss on thickness was also studied. Table

4.5 shows the comparison of dielectric constant and loss for films of different

thickness both measured at 10 kHz.

Table 4.5. Change in dielectric properties with increasing film thickness, for

different compositions.

Composition K (2 layers,

400 nm)

K (3 layers,

600 nm)

tan� (2 layers,

400 nm)

tan� (3 layers,

600 nm)

Ba0.5Sr0.5TiO3 105 213 0,07 0,05

Ba0.7Sr0.3TiO3 152 275 0,08 0,08

Ba0.9Sr0.1TiO3 156 213 0,08 0,05

The increase of dielectric constant with film thickness was obvious. This increase in

the dielectric constant with increasing thickness was expected because of three main

reasons: First, dielectric constant depends on grain size [7], and the grain size of the

films increases with increasing film thickness. Secondly, as the film thickness

increases, the density of dipoles increases, which in turn increases the polarization

density and dielectric constant. Third and most important reason is the presence of a

low dielectric constant interfacial layer between Pt bottom electrode and BST film

[43, 47]. In a thinner film near either electrode-film interface, the ionic contribution

to dielectric constant is suppressed; that is, a region of few lattice constant acts as a

low permittivity or a “dead” layer. An applied voltage is dropped across the

interfacial layers and bulk of the film in series. This effect of interfacial layer is more

pronounced for thinner films [17]. This low dielectric constant interfacial layer acts

as a series capacitor with the BST film and decreases the overall capacitance of BST

stack. Since the overall capacitance of these capacitors connected in series is equal to

78

1/Ceq = 1/Ci + 1/Cb, where Ceq is the overall capacitance, Ci is the capacitance of the

interfacial layer, and Cb is the bulk capacitance corresponding to the actual

capacitance of the BST film, interface layer with the low capacitance significantly

decreases the overall capacitance. This explains why bulk permittivities of BST,

which are in the order of thousands, could not be realized in thin films together with

grain size and thickness dependencies of thin films which are called together as size

effects in thin films. There are many suggestions explaining the possible reasons for

the formation of this interfacial layer [42], existence of which was also verified by

TEM recently [47].

Apart from grain size, dielectric constant of the films also depend on grain

morphology. Columnar-grained films are reported to exhibit larger dielectric

constants [37]. The change in magnitude can be dramatic: from 500 to 900. A

granular microstructure with random orientation of the grains typically occurs from

homogeneous nucleation in an amorphous matrix as obtained after the pyrolysis step

during chemical solution deposition (CSD). Often, there is a competition between

heterogeneous nucleation at the bottom interface which may lead to formation of

oriented or epitaxial films and random homogeneous nucleation. The nucleation type

is controlled by first, the film material, second, the type of substrate interface, and

third, the processing parameters. Columnar grain growth in thin films can only be

achieved if the possibility for a single nucleation event is favored over the other

nucleation events. Regarding to the theory of glass crystallization, heterogeneous

nucleation can be favored over homogeneous nucleation by a lowering of the driving

force for crystallization which can be achieved by increasing the crystallization

temperature or by the influence of an intermediate phase. In BST thin films columnar

grains were reported to be obtained only for precursor solutions which transform to

the perovskite phase via intermediate alkaline earth-titanium-oxo-carbonate phase.

Due to delayed crystallization, the driving force is lowered which results in a

favoring of heterogeneous nucleation events. In contrast to the intensively

investigated PZT thin films which crystallize via the intermediate fluorite phase and

easily form columnar grains, the BST films only exhibit a columnar grain growth for

a small thickness (approximately 10 nm) of the individual crystallized layers [37].

79

Dielectric properties also depend on the substrate – film interface. RTA (Rapid

Thermal Annealing) is a new method that is said to improve this interface [48, 49].

The great advantages of this method are a short annealing time and its relative

process simplicity as compared with the conventional furnace annealing. The

objective of the short processing time is to basically reduce the time and temperature

product such that the desired physical or chemical processes are completed while the

unwanted processes such as dopant diffusion, interface reactions, decomposition, or

evaporation, etc., are suppressed or minimized. The short rise time to the desired

annealing temperature minimizes the interface reactions and results in a dense

microstructure [48].

Ren et al. [38] synthesized BST films with the same procedure as in our study, only

difference being the usage of RTA instead of conventional annealing. They measured

a dielectric constant of 230 and dielectric loss of 0.02 for a 160 nm film, which is

better than the results in this study. They claim that RTA minimizes the formation of

low-dielectric layer between BST and substrate. A summary of the dielectric

properties of all samples produced in our study is given at Table 4.6.

As stated in Chapter 2, hillocks are formed in Pt/Ti electrodes to relieve the

compressive stress of Pt film [22]. Therefore hillock formation during BST film

formation was also investigated. For this purpose, surfaces of three substrates with

different thermal histories were inspected using SEM without coating the BST film

on top, in order to see if any metal protrudes from the surface. One of the substrates

was inspected in as-received condition without any thermal treatment, another was

annealed at 800˚C for 1 hours and last one annealed under the same conditions with

BST films, at 800˚C for 3 hours. SEM micrographs of three samples are given

respectively in Figure 4.23.

80

Table 4.4. Dielectric Properties of BaxSr1-xTiO3 films.

1 kHz 100 kHz 1 MHz Film Composition

Film Thickness

Heat Treatment Applied

Dielectric Constant

Dielectric Loss (%)

Dielectric Constant

Dielectric Loss (%)

Dielectric Constant

Dielectric Loss (%)

Ba0.5Sr0.5TiO3 600 nm 800˚C, 3 hrs 230 6.1 205 4.6 196 8.53

Ba0.5Sr0.5TiO3 600 nm 700˚C, 3 hrs 184 8,2 155 4,9 150 6,9

Ba0.5Sr0.5TiO3 600 nm 600˚C, 3 hrs 80 6.2 70 4.1 66 4.2

Ba0.5Sr0.5TiO3 400 nm 800˚C, 3 hrs 135 8.3 114 5.7 107 13.5

Ba0.9Sr0.1TiO3 600 nm 800˚C, 3 hrs 228 6.5 202 6.5 184 28.8

Ba0.9Sr0.1TiO3 400 nm 800˚C, 3 hrs 175 9.9 141 17 100 42.3

Ba0.7Sr0.3TiO3 600 nm 800˚C, 3 hrs 313 10.8 249 9.2 217 30.6

Ba0.7Sr0.3TiO3 400 nm 800˚C, 3 hrs 176 11,7 137 7,3 126 14,2

Ba0.5Sr0.5TiO3 (Old

Solution) 500 nm 800˚C, 3

hrs 188 7,83 160 4,8 152 6,9

81

Figure 4.23. SEM micrograph of the surfaces of Pt/Ti/SiO2/Si substrate: (top) prior

to any thermal treatment, (middle) after annealing at 800˚C for 1 hour. White dots

are hillocks of Pt, (bottom) after annealing at 800˚C for 3 hours. White dots have

grown in size.

82

Surface of as received Pt substrate was homogeneous and free of any different

contrast features. However, the substrate annealed at 800˚C for 1 hour contained

some light color dot-shaped features. For the substrate annealed at 800˚C for 3 hours,

these features grew in size. Same kind of dot-like features protruding from the

surface were also observed by Park et al. for a different Pt/Ti/SiO2/Si substrate [50].

These features were suspected to be hillocks, but first EDS analysis should have been

made to assure that they are not second phase particles formed as a result of diffusion

process.

The EDS analysis showed that these features are rich in Pt, which makes sense since

bottom layers of Ti and Si contributes less to the hillock regions compared to normal

regions.

Hillocks are known to form capacitor shorts or result in degradation of electrical

properties. In this study, the formation of hillocks was proved by SEM studies on

uncoated substrates. But in the actual case, the substrates went into thermal treatment

together with the BST layer coated on top of them. So, the formation of hillocks may

be enchanced or prevented by the presence of BST layer. Al-Shareef et al. [51]

studied the effect of PZT film on the formation of hillocks and found that the

presence of PZT film suppresses hillock formation. This suppression was explained

by (a) limited surface mobility of Pt atoms due to the presence of top layer, (b)

reduced oxygen diffusion to the Pt – Ti interface, (c) reduced stress in the Pt/Ti layer

due to the presence of PZT overlayer prior to annealing. On the other hand, Al-

Shareef et al. [51] and Nam et al. [22] experimentally verified that hillocks could

result in capacitor shorts. They observed that the increase in size of hillocks rather

than their density influences the shorting. In our study, some of BST capacitors

shorted, but it is hard to prove that shorts arise due to hillocks in Pt/Ti/SiO2/Si

substrates.

83

4.6. Ferroelectric Properties of BST Films

Change of capacitance with voltage, C-V behaviour in short, gives an idea about

ferroelectricity. Therefore, C-V measurements are mentioned within the ferroelectric

properties section. The polarization and capacitance vary non-linearly with an

applied field due to domain structure. The C-V profile is actually a measure of

dP/dE, so the peak in the C-V characteristic corresponds to a large polarization

change, or the switching of ferroelectric domains from one orientation to another.

The dc bias was increased from 0 V to 4.5 V in 0.5 V intervals, and then decreased

again to 0 V and increased in the reverse polarity down to –4.5 V. The C-V curve for

Ba0.7Sr0.3TiO3 is shown in Figure 4.24. The C-V curve was symmetric and nearly no

hysteresis was observed due to paraelectric nature and fine grain size of the film, as

expected for this composition which is above its Curie temperature at room

temperature. C-V curve for Ba0.9Sr0.1TiO3 is shown in Figure 4.25, which is in

ferroelectric state at room temperature. This time in contrast to the Ba0.7Sr0.3TiO3

film, there was a butterfly shape in the curve, suggesting a possible ferroelectricity in

the film, since BST of this composition is in ferroelectric phase at room temperature.

Ferroelectricity in BST thin films is suppressed. Generally for ferroelectric films

below certain grain size called as critical grain size, despite the material is in its

ferroelectric state (below its Curie temperature at the ambient temperature),

hysteresis loop and dielectric anomaly around the Curie temperature cannot be

observed. For barium titanate thin films, this critical grain size for the existence of

the ferroelectricity is around 120 nm [36]. The reason for disappearance of

ferroelectricity below this critical grain size is unknown [7]. Also, for BST thin films

Sr addition to BaTiO3 decreases the tetragonality (c/a ratio) in the perovskite

structure and a small value of c/a ratio may not be sufficient to create charge

separation and spontaneous polarization in the films, so the fabrication of BST

ferroelectric film is even more difficult than that of BaTiO3 [36].

84

C-V for 3 layers 0.7-0.3 film

880

884

888

892

896

900

-6 -4 -2 0 2 4 6

Voltage (V)

Cap

acita

nce

(pF)

Figure 4.24. C-V curve for a 3 layers Ba0.7Sr0.3TiO3 film.

C-V for 3 layers 0.9-0.1 film

1360

1380

1400

1420

1440

1460

1480

-3 -2 -1 0 1 2 3

Voltage (V)

Cap

acita

nce

(pF)

Figure 4.25. C-V curve for a 3 layers Ba0.9Sr0.1TiO3 film.

85

In this study, conventional sol-gel technique was used. Grain sizes of the films were

small, obtained from very rough approximation of Scherrer’s formula as 25 nm. This

grain size is very small with respect to the critical grain size reported for BT films,

which is 120 nm [36]. However, slim hysteresis loops were obtained by the

ferroelectric measurements made by constructing modified Sawyer-Tower circuit.

Among all BST films produced, only for 3 layers film (600 nm thickness) of

composition Ba0.9Sr0.1TiO3, hysteresis loops could be obtained. In figure 4.26,

hysteresis of this film obtained at 3 V peak to peak voltage is shown. For other

compositions, Ba0.7Sr0.3TiO3 and Ba0.5Sr0.5TiO3, polarization was linear with applied

electric field. This was expected because for bulk BST, 0.3 % Sr addition brings the

Curie temperature to room temperature, so for Ba0.7Sr0.3TiO3 and Ba0.5Sr0.5TiO3 the

material is in paraelectric state at room temperature, likewise C-V curve of

Ba0.7Sr0.3TiO3 did not display butterfly shape unlike Ba0.9Sr0.1TiO3. Hysteresis loop

parameters for the Ba0.9Sr0.1TiO3 film are given in Table 4.7. It can be seen that both

remnant polarization and coercive field values increase with increasing electric field.

This is due to alignment of domains with the field. Actual hysteresis loops captured

directly from the oscilloscope screen are given in appendix.

Hysteresis Loop for BST (0,9-0,1) film

-0,4

-0,3

-0,2

-0,1

0

0,1

0,2

0,3

0,4

-80 -60 -40 -20 0 20 40 60 80

Electric Field (kV/cm)

Pol

ariz

atio

n (�

coul

omb/

cm2)

Figure 4.26. Polarization vs electric field behaviour for 600 nm Ba0.9Sr0.1TiO3 film.

86

Table 4.7. Ferroelectric properties of Ba0.9Sr0.1TiO3 films on Pt/Ti/SiO2/Si

substrates.

Film Thickness

Film Composition

Coercive Field, Ec, (kV/cm)

Remanent Polarization, Pr (�C/cm2)

Applied E- Field (Emax)

(kV/cm)

Applied Voltage (Vmax)

4,5 0.0235 41,6 2,5 4,6 0,0313 50 3 0,6 �m Ba0.9Sr0.1TiO3 7,23 0,047 58,3 3,5 V

Remnant polarization and coercive field values of the films are very low for a

ferroelectric thin film. This is due to fine grain size of the films. The sol-gel

deposition of BaTiO3 films commonly results in polycrystalline, granular films with

grain diameters of lower than 70 nm due to random nucleation in the pyrolyzed gel

films [36]. Recently, some researchers [36, 52] managed to fabricate ferroelectric

BST films with grain sizes greater than 100 nm by coating highly diluted solutions

on lattice and structure matched substrates and sintering each coated layer by rapid

thermal annealing in order to promote epitaxial, oriented growth by heterogeneous

nucleation from the substrate. For example, Wang et al. [52] succeeded at fabricating

columnar grained, 250 nm thick Ba0.9Sr0.1TiO3 films having grain sizes in 80-200 nm

range by coating 0.05M solutions on Pt/Ti/SiO2/Si substrates. They reported remnant

field of 7.8 �C/cm2 and coercive field of 56 kV/cm at an electric field of 320 kV/cm.

87

CHAPTER 5

CONCLUSIONS AND FURTHER SUGGESTIONS

In this study, stable (Ba,Sr)TiO3 solutions were synthesized by chemical solution

deposition, and coated on Pt/Ti/SiO2/Si and Si substrates with different processing

parameters like solution concentration, precursor materials, spinning speed, heat

treatment temperatures, and number of layers coated. The microstructure change

upon change of these parameters was related to the measured dielectric properties.

It was found that addition of ethylene glycol stabilizes the solution and helps the

dissolution of acetates.

Crack-free, homogeneous and dense BaxSr1-xTiO3 films were obtained. Optimum

spinning speed and time was determined as 4000 rpm, 30 seconds, whereas x-ray

diffraction and dielectric constant measurements suggested optimum sintering

parameters as 800˚C for 3 hours for the complete crystallization of the films.

Dielectric constant of the films increased with increasing temperature, in agreement

with the crystallization studies. It increased from 80 to 230 for a 600 nm

Ba0.5Sr0.5TiO3 film at 1 kHz, when the sintering temperature increased from 600˚C to

800˚C.

Dielectric constant of the films was slightly decreased with frequency in the range 1

kHz – 1 MHz.

88

Dielectric constant of the films also changed with Ba/Sr ratio, but this change was

not linear with the increase in Sr content. Dielectric constant of Ba0.5Sr0.5TiO3 and

Ba0.9Sr0.1TiO3 was close to eachother. For 600 nm film thickness, the former was 230

and the latter was 228 at 1 kHz. Dielectric constant of Ba0.7Sr0.3TiO3 was the highest;

it was 313 for the same thickness with others, It is known that 0.3 mole percent Sr

brings Curie temperature of pure BaTiO3 to room temperature and dielectric constant

displays a peak near phase transition temperature.

Dielectric loss of the films was changed between 4 – 9 % in the measurement

frequency range 1kHz – 100 kHz, but increased sharply above 100 kHz up to 30 %

similar to what has been reported in the literature.

Dielectric constant increased but the dielectric loss decreased with increasing film

thickness for all compositions. The increase in the dielectric constant was due to the

minimization of the effect of low dielectric constant interfacial layer in series with

the film with increasing film thickness, the reduction in the fraction of pores due to

increased number of coating cycles and the increase in the dipole moment density

with thickness. The decrease in dielectric loss might be due to reduction of resistive

loss component of the total dielectric loss although the dipole relaxation component

should increase with film thickness due to increase in the density of dipoles.

The BST films showed typical C-V curves, capacitance decreased with increasing

bias voltage. For the Ba0.5Sr0.5TiO3 film, which is above Curie temperature at room

temperature, there was no change in C-V curve for negative and positive polarity,

suggesting that the film is paraelectric. On the contrary, Ba0.9Sr0.1TiO3 films had a

slightly butterfly shaped C-V curve, suggesting a possible ferroelectricity in the film.

But, although films with that composition are ferroelectric in bulk form,

polycrystalline BST thin films with small equiaxed grains fails to show

ferroelectricity.

89

Only, BST films with the composition Ba0.9Sr0.1TiO3 displayed hysteresis. Remnant

polarization of the film was 0.047 �C/cm2 and coercive field was 7.23 kV/cm at an

applied field of 58 kV/cm.

For the improvement of the dielectric properties, clean room environment for the

preparation of the substrate is a must. Also, RTA (rapid thermal annealing) might be

useful in reducing the interfacial layer formed between the film and the substrate.

Oriented or epitaxial films can be synthesized by coating thinner layers using highly

diluted solutions on structure and lattice matched oxide substrates like LaAlO3 or

MgO and applying RTA. The usage of oxide electrodes limits the dielectric losses.

Measurement of leakage current of the films may be helpful to characterize the

electrical properties. Leakage current is a crucial parameter for DRAM applications.

Increasing sintering temperature improves the densification by decreasing the pore

density. This increase in densification can be monitored by measuring refractive

index of the films using spectroscopic ellipsometry or visible-UV spectroscopy.

The dielectric properties strongly depend on grain size and morphology of the films.

Since, small equiaxed grains of BST films synthesized by sol-gel processing can not

be differentiated by SEM, higher magnification electron microscopes like TEM or

FESEM may be used to measure grain size.

90

REFERENCES

[1] N. Setter, R.Waser, “Electroceramic Materials”, Acta Materialia, Vol. 48, pp.

151-178, (2000).

[2] A.I. Kingon, S.K. Streiffer, “Ferroelectric Films and Devices”, Current

Opinion in Solid State & Materials Science, Vol. 4, (1999).

[3] International Technology Roadmap for Semiconductors (ITRS) Report

(2001), http://public.itrs.net.

[4] Principles of Electronic Ceramics, L.L. Hench and J.K. West, John Wiley &

Sons, (1989).

[5] Ceramic Materials for Electronics, Edited by R. C. Buchanan, Marcel Dekker

Inc., (1991).

[6] Cem Ba�çeri, “Electrical and Dielectric Properties of (Ba,Sr)TiO3 Capacitors

for Ultra-High Density Dynamic Random-Access Memories”, PhD

Dissertation, North Carolina State University, (1997).

[7] Dragan Damjanovic, “Ferroelectric, Dielectric and Piezoelectric Properties of

Ferroelectric Thin Films and Ceramics”, Rep. Prog. Phys. 61, pp 1267-1324,

(1998).

[8] G.H. Haertling, “Ferroelectric Ceramics: History and Technology”, JACS,

Vol. 82 [4], pp. 797-818, (1999).

91

[9] B.A. Baumert, L.H. Chang, A.T. Matsuda, C.J. Tracy, N.G. Cave, R.B.

Gregory, P.L. Fejes, “A Study of Barium Strontium Titanate Thin Films for

Use in Bypass Capacitors”, Journal of Materials Research, Vol. 13 [1], pp.

197-204, (1998).

[10] S. Ezhilvalavan, T.Y. Tseng, “Progress in the Developments of (Ba,Sr)TiO3

(BST) Thin Films for Gigabit Era DRAMs”, Materials Chemistry and

Physics, Vol.65, pp. 227-248, (2000).

[11] Piezoelectric Ceramics, Bernard Jaffe, Academic Press, (1971).

[12] A. I. Kingon, J.P. Maria, S.K.Streiffer, “Alternative Dielectrics to Silicon

Dioxide for Memory and Logic Devices”, Nature, Vol. 406, pp. 1032-1038,

(2000).

[13] D.M. Tahan, A. Safari, L.C. Klein, “Sol-Gel Preparation of Barium Strontium

Titanate Thin Films”, JACS, Vol. 79, pp. 1593, (1996).

[14] J.G. Cheng, X.J. Meng, J. Tang, S.L. Guo, J.H. Chu, “Fabrication and

Electrical Properties of Sol-Gel Derived Ba0.8Sr0.2TiO3 Ferroelectric Films

from a 0.05-M Spin-On Solution”, Applied Physics A, Vol. 70, pp. 411-414,

(2000).

[15] S. Liu, M. Liu, Y. Zeng, Y. Huang, “Preparation, Phase Transition

Diffuseness and Ferroelectric Fatigue Behaviour of Ba0.85Sr0.15TiO3 Thin

Film for Uncooled Infrared Focal Plane Arrays”, Materials Science and

Engineering B, Vol. 90, pp. 149-153, (2002).

[16] S.B. Majumder, M. Jain, A. Martinez, R.S. Katiyar, F.W. Van Keuls, F.A.

Miranda, “Sol-Gel Derived Grain Oriented Barium Strontium Titanate Thin

Films for Phase Shifter Applications”, Journal of Applied Physics, Vol.90

[2], pp. 896-903, (2001).

92

[17] T.M. Shaw, Z. Suo, M. Huang, E. Liniger, R.B. Laibowitz, J.D. Baniecki,

“The Effect of Stress on the Dielectric Properties of Barium Strontium

Titanate Thin Films”, Applied Physics Letters, Vol. 75 [14], pp. 2129-2131,

(1999).

[18] R.W. Schwartz, P.G. Clem, J.A. Voight, E.R. Byhoff, M.V. Stry, T.J.

Headley, N.A. Missert, “Control of Microstructure and Orientation in

Solution-Deposited BaTiO3 And SrTiO3 Thin Films”, JACS, Vol. 82 [9], pp.

2359-2367, (1999).

[19] M. Nayak, S.Y.Lee, Tseung-Yuen Tseng, “Electrical and Dielectric

Properties Of (Ba0.5Sr0.5)TiO3 Thin Films Prepared by a Hydroxide-Alkoxide

Precursor-Based Sol-Gel Method”, Materials Chemistry and Physics, Vol. 77,

pp. 34-42, (2002).

[20] J.M. Bell, P.C. Knight, G.R. Johnston, “Electrode Materials for Ferroelectric

Film Capacitors and Their Effect on The Electrical Properties”, Ferroelectric

Thin Flms: Synthesis and Basic Properties, edited by C.A. Paz de Araujo, J.F.

Scott, G.W. Taylor, Gordon and Breach Publications, (1996).

[21] H.N.Al-Shareef, A.I. Kingon, “Electrode Materials for Ferroelectric Film

Capacitors and Their Effect on the Electrical Properties”, Ferroelectric Thin

Flms: Synthesis and Basic Properties, edited by C.A. Paz de Araujo, J.F.

Scott, G.W. Taylor, pp. 193-230, Gordon and Breach Publications, (1996).

[22] H.J. Nam, D.K. Choi, W.J. Lee, “Formation of Hillocks in Pt/Ti Electrodes

and Their Effects on Short Phenomena of PZT Films Deposited by Reactive

Sputtering”, Thin Solid Films, Vol. 371, pp. 264-271, (2001).

[23] M. C. Gust, L.A. Momoda, N.D. Evans, M.L. Mecartney, “Crystallization of

Sol-Gel Derived Barium Strontium Titanate Films”, JACS, Vol.84 [5], pp.

1087-92, (2001)

93

[24] Brian D. Fabes, Brian J. J. Zelinski, Ronald R. Uhlmann, “Sol-Gel Derived

Ceramic Coatings”, in Ceramic Films and Coatings, Edited by John B.

Wachtman, Richard A. Haber, Noyes Publications, Park Ridge, NJ, USA,

(1993)

[25] D. E. Bornside, C. W. Macosko, L. E. Scriven, Journal of Imaging

Technology, Vol. 13, pp. 122-129, (1987).

[26] Sol-Gel Science, C.J. Brinker, G.W. Scherer, Academic Press, (1990).

[27] G. Yi, M. Sayer, “Sol-Gel Processing of Complex Oxide Films”, Ceramic

Bulletin, Vol. 70, No. 7, pp. 1173-1179, (1991).

[28] E. Dien, R. Verveur, M. Lejeune, A. Smith, “Preparation and Properties of

Ba1-xSrxTiO3 Thin Films Deposited by the Sol-Gel Technique”, Journal of

Physics IV France, Vol. 8, pp. 9-73-9.77, (1998).

[29] F.M. Pontes, E.R. Leite, D.S.L. Pontes, E. Longo, E.M.S. Santos, S.

Mergulhao, P.S. Pizani, F. Lanciotti, T.M. Boschi, “Ferroelectric and Optical

Properties of Ba0.8Sr0.2TiO3 Thin Film”, Journal of Applied Physics, Vol. 91

[9], pp. 5972-5978, (2002).

[30] H.Y. Tian, W.G. Luo, X.H. Pu, A.L. Ding, “Synthesis and Microstructure of

the Acetate-Based Sr-Doped Barium Titanate Thin Films Using a Modified

Sol-Gel Technique”, Journal of Materials Science Letters, Vol. 19, pp.1211-

1213, (2000).

[31] U. Hasenkox, S .Hoffmann, R.Waser, “Influence of Precursor Chemistry on

the Formation of MTiO3 (M = Ba,Sr) Ceramic Thin Films”, Journal of Sol-

Gel Science and Technology, Vol.12, pp. 67-79, (1998).

94

[32] D.M. Tahan, “Microstructure and Property Characterization of Barium

Strontium Titanate Thin Films Prepared by a Sol-Gel Technique For DRAM

Applications”, Ph.D. Dissertation, Rutgers University, (1997).

[33] D.P. Birnie II, “Esterification Kinetics in Titanium Isopropoxide-Acetic Acid

Solutions” Journal of Materials Science, Vol. 35, pp. 367-374, (2000).

[34] L.L. Lopez, J. Portelles, J.M. Siqueiros, G.A. Hirata, J. McKittrick,

“Ba0.5Sr0.5TiO3 Deposited by PLD on SiO2/Si, RuO2/Si and Pt/Si Electrodes,

Thin Solid Films, Vol. 373, pp. 49-52, (2000).

[35] S. Kim, T.S. Kang, J.H. Je, “Structural Characterization of Laser Ablated

Epitaxial (Ba0.5Sr0.5)TiO3 Thin Films on MgO (001) by Synchotron X-ray

Scattering”, Journal of Materials Research, Vol.14[7], pp.2905, 1999.

[36] J.G. Cheng, X.J. Meng, J. Tang, S.L. Guo, J.H. Chu, “Fabrication and

Electrical Properties of Sol-Gel-Derived Ba0.8Sr0.2TiO3 Ferroelectric Thin

Films from a 0.05-M Spin-On Solution”, Applied Physics A, Vol. 70, pp.

411-414, (2000).

[37] S. Hoffmann, R. Waser, “Control of the Morphology of CSD-Prepared

(Ba,Sr)TiO3 Thin Films”, Journal of the European Ceramic Society, Vol.19,

pp. 1339-1343, (1999).

[38] T.L. Ren, X.N. Wang, J.S. Liu, H.J. Zhao, T.Q. Shao, L.T. Liu, Z.J. Li,

”Characteristics of Silicon-Based BaxSr1-xTiO3 Thin Films Prepared by a Sol-

Gel Method, Journal of Physics D: Applied Physics, Vol. 35, pp. 923-926,

(2002).

[39] Elements of X-ray Diffraction, B.D. Cullity, Addison-Wesley Publishing

Company, Inc., (1978).

95

[40] F.M. Pontes, E. Longo, E.R. Leite, J.A. Varela, “Study of the Dielectric and

Ferroelectric Properties of Chemically Processed BaxSr1-xTiO3 Thin Films”,

Thin Solid Films, Vol. 386, pp. 91-98, (2001).

[41] N.V. Giridharan, R. Jayavel, P. Ramasamy, “Structural, Morphological and

Electrical Studies on Barium Strontium Titanate Thin Films Prepared by Sol-

Gel Technique”, Crystal Research Technology, Vol. 36 [1], pp. 65-72,

(2001).

[42] S.K. Streiffer, C. Ba�çeri, C.B. Parker, S.E. Lash, A.I. Kingon,

”Ferroelectricity in Thin Films: The Dielectric Response of Fiber-Textured

(BaxSr1-x)Ti1+yO3+z Thin Films Grown by Chemical Vapor Deposition”,

Journal of Applied Physics, Vol.86 [8], (1999).

[43] S.Stemmer, S.K. Streiffer, N.D. Browning, C.Ba�çeri, A.I. Kingon, “Grain

Boundaries in Barium Strontium Titanate Thin Films: Structure, Chemistry

and Influence on Electronic Properties, Interface Science, Vol. 8, pp. 209-

221, (2000).

[44] Yang G, Gu H, Zhu J, Wang Y, “The Fabrication and Characteristics of

(Ba0.5Sr0.5)TiO3 Thin Films Prepared by Pulsed Laser Deposition”, Journal of

Crystal Growth, Vol. 242, pp. 172-176, (2002).

[45] Fitsilis F, Regnery S, Ehrhart P, Waser R, Schienle F, Schumacher M,

Dauelsberg M, Strzyzewski P, Juergensen H, “BST Thin Films Grown in a

Multiwafer MOCVD Reactor”, Journal of the European Ceramic Society,

Vol. 21, pp. 1547-1551, (2001).

[46] Im J, Auciello O.P.K., Baumann P.K., Streiffer S.K., Kaufman D.Y., Krauss

A.R., “Composition Control of Magnetron Sputter Deposited (BaxSr1-

x)Ti1+yO3+z Thin Films for Voltage Tunable Devices”, Applied Physics

Letters, Vol.76 [5], pp. 625-627, (2000).

96

[47] H.Z. Jin, J. Zhu, P.Ehrhart, F. Fitsilis, C.L. Jia, S. Regnery, K. Urban, R.

Waser, “An Interfacial Defect Layer Observed at (Ba,Sr)TiO3/Pt Interface”,

Thin Solid Films, Vol. 429, pp. 282-285, (2003).

[48] P.J. Joshi, S. Ramanathan, S.B. Desu, S. Stowell, “Microstructural and

Electrical Characteristics of Rapid Thermally Processed (Ba1-xSrxTiO3)

Prepared by Metalorganic Solution Deposition Technique”, Physica Status

Solidi (a), Vol. 161, pp. 361-370, (1997).

[49] D. Wu, A. Li, H. Ling, X. Yin, C. Ge, M. Wang, N. Ming, “Preparation of

(Ba0.5Sr0.5)Tio3 Thin Films by Sol-Gel Method with Rapid Thermal

Annealing, Applied Surface Science, Vol.165, pp. 309-314, (2000).

[50] K.H.Park, C.Y.Kim, Y.W.Jeong, H.J.Kwon, K.Y.Kim, J.S.Lee, S.T.Kim

“Microstructures and Interdiffusions of Pt/Ti Electrodes with respect to

Annealing in the Oxygen Ambient“, Journal of Materials Research, Vol.10

[7], pp. 1790-1794 (1995).

[51] H.N. Al-Shareef, D. Dimos, B.A. Tuttle, M.V. Raymond, “Metallization

Schemes for Dielectric Thin Film Capacitors”, Journal of Material Research,

Vol. 12 [2], pp. 347-354, (1997).

[52] G.S. Wang, J. Yu, Q. Wang, Q. Zhao, J.L. Sun, X.J. Meng, S.L. Guo, J.H.

Chu, “Structure and Optical Properties of Ba0.9Sr0.1TiO3 Ferroelectric Thin

Films Prepared by Chemical Solution Routes”, Physica Status Solidi (a), Vol.

194 [1], pp. 56-63, (2002).

[53] “Powder Diffraction File, Alphabetical Index: Inorganic Phases”, (JCPDS),

Swarthmore, PA, (1986).

97

APPENDIX

Actual Hysteresis Loops

Figure A.1. Polarization versus electric field behaviour for 600 nm Ba0.9Sr0.1TiO3

film at Vx = 3.5 V.

98

Figure A.2. Polarization versus electric field behaviour for 600 nm Ba0.9Sr0.1TiO3

film at Vx = 3 V.

Figure A.3. Polarization versus electric field behaviour for 600 nm Ba0.9Sr0.1TiO3

film at Vx = 2.5 V.